Second-Order Cybernetics

1 Eigenforms, Interfaces and

column A column B column C
1

Holographic Encoding
2 2
3 3
4 4
5 5
6
7
8
Toward an Evolutionary Account 6
7
8
9
10
of Objects and Spacetime 9
10
11
Chris Fields • Independent Scholar • fieldsres/at/gmail.com 11
12 12
13 Donald D. Hoffman • University of California, Irvine, USA • ddhoff/at/uci.edu 13
14
Chetan Prakash • California State University, USA • cprakash/at/csusb.edu 14
15 15
Robert Prentner • ETH Zurich, Switzerland • robert.prentner/at/phil.gess.ethz.ch
Physics Concepts in Second-Order Cybernetics

16 16
17 17
18 > Context •The evolution of perceptual systems and hence of observers remains largely disconnected from the ques- 18
19 tion of the emergence of classical objects and spacetime. This disconnection between the biosciences and physics im- 19
20 pedes progress toward understanding the role of the “observer” in physical theory. > Problem • In this article we con- 20
21 sider the problem of how to understand objects and spacetime in observer-relative evolutionary terms. > Method • We 21
22 rely on a comparative analysis using multiple formal frameworks. > Results • The eigenform construct of von Foerster 22
23 is compared to other formal representations of observer–environment interactions. Eigenforms are shown to be en- 23
24 coded on observer-environment interfaces and to encode fitness consequences of actions. Space and time are com- 24
25 ponents of observational outcomes in this framework; it is suggested that spacetime constitutes an error-correcting 25
26 code for fitness consequences. > Implications • Our results contribute to an understanding of the world in which nei- 26
27 ther objects nor spacetime are observer-independent. > Constructivist content • The eigenform concept of von Foer- 27
28 ster is linked to the concepts of decoherence and holographic encoding from physics and the concept of fitness from 28
29 evolutionary biology. > Key words • Active inference, boundary, conscious agent, icon, Markov blanket, redundancy. 29
30 30
31 31
32 32
33 33
34 Introduction such self-reproducing eigenform-eigenbe- or Howard Pattee (2001) between agent and 34
35 havior systems to a principle of cosmology: world for the purposes of theory construc- 35
36 « 1 » Heinz von Foerster (1976) intro- tion. It is from this perspective that an ei- 36
200 37 duced the eigenform and eigenbehavior
38 concepts by considering an agent that both
“ The Universe is constructed in such a way that
it can refer to itself […] the universe can pretend
genform becomes, or perhaps better, serves
as an object that the agent observes and acts
37
38
39 observes and acts on a surrounding world: that it is two and then let itself refer to the two, with respect to. This agent-centered per- 39
40 an eigenform is an observation that remains and find that it has in the process referred only spective, when combined with the essential 40
41 invariant, in the limit of long interaction
42 time, under some class of behaviors, while
”
to the one, that is, itself. (Kauffman 2009: 134) external perspective of the theorist, allows
us to consider the ecological situation of an
41
42
43 an eigenbehavior is an action that, in the This formulation makes explicit an impor- agent for whom every observation presents 43
44 same limit, leaves some eigenform invari- tant point: that there is no difference in sub- multiple objects, every object allows mul- 44
45 ant. These concepts naturally suggest an stance, and hence no metaphysical dualism, tiple actions, and every pairing of an object 45
46 abstract picture in which the eigenbehavior between agent and environment. with an action has consequences that may 46
47 continually reproduces the eigenform, inde- « 2 » Here we pursue the notion of an ei- be good or bad for the agent. We compare 47
48 pendently of any other features or dynamics genform not from the perspective of an ab- the description of this situation in terms of 48
49 of the world. In this picture, eigenform and stract reflexive system, but rather from von eigenforms to its description in two inde- 49
50 eigenbehavior compose a single reflexive Foerster’s original perspective of an agent pendently developed formal representations 50
51 system; all other aspects of the world can that observes and acts on its world, a world of the agent-world interaction: the con- 51
52 be neglected. Louis Kauffman has shown, that can be taken to be the rest of the Uni- scious agent formalism of Donald Hoffman 52
53 conversely, that all such reflexive systems verse in which the agent is embedded. We and Chetan Prakash (2014) and the Markov 53
54 have eigenforms and eigenbehaviors as in- impose, in other words, an “epistemic cut” in blanket formalism of Judea Pearl (1988) as 54
55 variants. Kauffman elevates the reflexivity of the sense used by John von Neumann (1955) applied to biological systems by Karl Friston 55
column A column B column C

Constructivist Foundations vol. 12, N°3

 Second-Order Cybernetics
TOC Chris Fields et al.

column A column B column C

1 (a) (b) (c) (d) 1
2 Obs2 A1 = P2 2
G1 X2
3 Alice Bob 3
4 S2 Alice Bob D1 D2 4
S1
5 5
X1 G2 Markov
6 Obs1 6
Boundary S P1 = A2 blanket
7 7
8 8
9 Figure 1 • Four representations of two-agent, or alternatively, agent-environment interaction. (a) Two agents S1 and S2, here depicted as compu- 9
10 tational processes, exchange observations Obs1 and Obs2 (adapted from Foerster 1976: 94). (b) Two agents, or alternatively two classical black 10
11 boxes, Alice and Bob exchange inputs and outputs across a boundary S that is in principle arbitrarily movable as described in Fields (2016). Alice’s 11
12 outputs are Bob’s inputs and vice versa. (c) Two conscious agents as defined by Hoffman and Prakash (2014) act on each other. Here X1 and G1 12
13 and X2 and G2 are measurable spaces representing the experiences and available actions, respectively, of the two agents; D1 and D2, P1 and P2, and 13
14 A2 and A2 are Markov kernels representing the decision processes, perceptions, and executed actions, respectively, of the two agents. 14
15 (d) Two agents interact via an intervening Markov blanket as described in Friston (2013). Arrows represent Markov processes. 15
16 16
17 17
18 (2013). In both of these latter representa- mation not just about its own stability, but of a theorist describing the overall situation 18
19 tions, the agent’s observations and actions also about the stability of other eigenforms. “from the outside” that the two agents and 19
20 “pass through” a boundary or interface that What kind of encoding, we then ask, can their exchange of observations within the 20
21 separates the agent – even if this separation have this property? We suggest that spacet- closed-loop system can be made explicit. 21
22 is purely notional – from its observed world. ime itself, including both the space in which « 5 » The closed-loop, two-agent ex- 22
23 We show that eigenforms can be regarded as objects appear to be embedded and the time change in Figure 1a involves an apparent par- 23
24 “icons” specifying possible interactions that over which they appear to persist, is a rela- adox: each agent receives information from 24
25 are encoded on this interface. We then sug- tional, error-correcting code for the fitness the other, so the total information in a two- 25
26 gest that this notion of an encoding of in- consequences of interactions. The forms agent system appears to increase. Any such 26
27 formation about possible interactions on an and locations of “objects” in “space” encode increase in a closed system, as von Foerster 27
28 interface is in fact very general, by showing probabilistic information about what future (1960) notes, appears to violate the 2nd law 28
29 that it corresponds to the notion of holog- interactions with these or other objects, if of thermodynamics. Indeed, any agent, as a 29
30 raphy developed within quantum informa- they occur at all, may be like. The persis- self-organizing system, must “eat energy and 30
31 tion theory. In this case, the encoding can tence of an “object” in “time” encodes the order from its environment” (Foerster 1960: 31
32 be regarded as “recorded” by the process of robustness of the corresponding eigenform 36) in order to survive; from the perspective 32
33 quantum decoherence, confirming the close as an attractor. Eigenforms have evolved, we of any such agent, the order in its environ- 33
34 relationship between the eigenform concept argue, to make this encoding of future con- ment must decrease as it is “eaten.” The en- 34
35 and quantum theory already suggested by sequences as precise as possible given the vironment of either agent in Figure 1a is the 35
36 Kauffman (2003, 2011). energetic and other resource constraints of other agent; hence each agent must perceive 36
37 « 3 » Considering eigenforms as encod- the encoding interface. the other as losing information. It is here 37 201
38 ings of information for a particular agent that the difference between the agents’ and 38
39 on that agent’s interface with its observed the theorist’s perspectives becomes criti- 39
40 world allows us to ask what information an The interface cally important. As Max Tegmark (2012) 40
41 eigenform encodes. If perceived “objects” remarks in a similar context, neither agent 41
42 are tokens for eigenforms, what is their in- « 4 » As von Foerster recognized, a re- has observational access to the total entropy 42
43 formational role? The Interface Theory of flexive model escapes solipsism when the of the two-agent system (neither agent has 43
44 Perception (ITP) of Hoffman, Manish Singh “world” or “environment” of each agent the theorist’s perspective); neither agent can 44
45 and Prakash (2015) provides a prima facie includes other agents, or in the limit is an- get “outside” the system to measure the total 45
46 surprising answer: that “objects” do not en- other agent (e.g., Foerster 1960). Such a entropy. The total entropy of the two-agent 46
47 code information about the ontological or two-agent model is shown in Figure 1a; here system could be zero, as indeed it would be 47
48 causal structure of the world, but rather in- two agents S1 and S2 exchange observations if the agents were quantum-mechanical sys- 48
49 formation about the structure of the fitness Obs1 and Obs2 (Foerster 1976: 94). From the tems with an entangled joint state (in this 49
50 function that relates the agent to the world. perspective of either agent, the other agent case, each agent would see itself communi- 50
51 This information is object-relative, but not is its entire “world” and every observation cating, but an outside observer would see no 51
52 object-specific: an interaction with one appears to be an observation of this entire communication as discussed further below). 52
53 object can have fitness consequences that world; there is nothing else with which the It is only the agents’ principled lack of obser- 53
54 affect interactions with other objects. An agent interacts, and hence nothing else that vational access to the system in which they 54
55 eigenform, in other words, encodes infor- it observes. It is only from the perspective are embedded that allows each agent to con- 55
column A column B column C

http://constructivist.info/12/3/200.fields
 column A column B column C
1 sider itself to be gaining information at the (Fields 2016a), and is hence what allows the not change when the “right” action – the 1
2 expense of its environment. Hence the sec- two-agent representation in Figure 1a. eigenbehavior – is executed. Such an ei- 2
3 ond law is respected from each agent’s indi- « 7 » The position of the boundary S genpercept has persistence over time if the 3
4 vidual perspective. This comports well with separating Alice from Bob in Figure 1b is, right action is taken; the wrong action may 4
5 the probabilities that appear in the second like the total entropy of the joint Alice + Bob lead to its disappearance. An autonomous 5
6 law’s being subjective, not objective. system, definable only from the “god’s eye” agent must choose the right action to take in 6
7 « 6 » The lack of observational access perspective of the theorist. Moving the any particular circumstance, i.e., given any 7
8 that rescues Figure 1a from paradox has a boundary changes the “sizes” of Alice and combination of current state and current 8
9 second important consequence: the envi- Bob and hence their definitions as “sys- percept. To the eigenform-eigenbehavior 9
10 ronment of each agent becomes a classical tems.” It also changes what “counts” for each concept, therefore, we may add the notion 10
11 black box, a system to which observers have of them as an input or an output. However, of an eigendecision, the decision to execute 11
12 only external access. More formally, a clas- moving the boundary S changes nothing the eigenbehavior that results in renewal 12
13 sical black box is a system about which no about the relationship of mutual exchange of the eigenform. While autonomy in the 13
14 observer can have more (non-hypothetical) between Alice and Bob, and indeed nothing non-trivial machine sense inferred above 14
15 information than is contained in a finite about the behavior of the joint system they is somewhat abstract, a requirement for au- 15
Physics Concepts in Second-Order Cybernetics

16 list of finite-length bit strings representing compose. This invariance under changes in tonomous decision-making at least suggests 16
17 observed input-output transitions (Ashby the positions of boundaries drawn by theo- an awareness of potential consequences and 17
18 1956; for a recent review, see Fields 2016a). rists is built deeply into the formalisms of hence consciousness. 18
19 Because neither agent can see “inside” the both classical and quantum physics (Fields « 9 » A minimal formal model of a con- 19
20 black box of its environment – this is, after 2016a); it is, indeed, this invariance that al- scious agent (CA) that experiences percep- 20
21 all, what “no observational access” means – lows theorists to choose “systems of inter- tual input from the world W in which it is 21
22 neither agent knows what its environment est” arbitrarily. It is implicit in von Foerster’s embedded, decides between possible actions 22
23 contains. The two agents of Figure 1a can, (1976) and Kauffman’s (2009) reduction of to take on the basis of that input, and then 23
24 therefore, also be represented as two inter- the agent-environment dynamics to the re- executes the selected action on W has been 24
25 acting black boxes; we give them their tra- flexive dynamics of a single, unitary system. developed by Hoffman and Prakash (2014), 25
26 ditional names Alice and Bob (Figure 1b; cf. The Alice–Bob boundary being arbitrarily who show that this minimal model is com- 26
27 the similar construction of Ranulph Glan- movable means that Alice and Bob do not putationally universal. They propose as the 27
28 ville 1982: Figure 5, where the theorist’s per- know, and cannot determine, where in the thesis of “conscious realism” that the world 28
29 spective is made explicit). Alice gives inputs joint system their mutual boundary is. Each W can always be considered to itself be a 29
30 to the unknown system Bob and receives can only locate the boundary from her or his CA; in this case, the agent-world interaction 30
31 outputs in return; the situation is the same own perspective; the “god’s eye” perspective can be represented as in Figure 1c (adapted 31
32 from Bob’s point of view. Edward Moore’s needed to locate it within the joint system from Hoffman and Prakash 2014: Figure 2). 32
33 (1956) theorem assures that neither Alice is unavailable. Not only can they not ob- Conscious realism incorporates, clearly, the 33
34 nor Bob can determine the complete state serve the “interior” of their interaction part- assumption discussed above that the limit in 34
35 space or dynamics of the other from finite ner/environment, they cannot observe the which the other agent “fills” the entire en- 35
36 input-output observations (see Fields 2013, boundary separating themselves from their vironment exists. As in the case of a black- 36
202 37 2016 for extensive discussion). Either must, partner/environment. All that either Alice box agent, this assumption can be stated as 37
38 therefore, regard the other as a “non-trivial or Bob can observe is the sequence of “in- a claim about observational access: no agent 38
39 machine,” i.e., as a system whose behavior is puts” that cross their respective boundaries can demonstrate by observation that its en- 39
40 unpredictable in principle as von Foerster from their respective environments. These vironment or any component thereof is not 40
41 (1973) emphasizes. Principled unpredict- sequences of inputs are the totality of their also a conscious agent. Conscious realism 41
42 ability is considered by some to indicate perceptual, as opposed to internally gener- makes each agent’s action the other agent’s 42
43 autonomy or “free will” and hence agency ated or introspective, experiences. perception in Figure 1c, just as they are in 43
44 from the perspective of external observers « 8 » As agents, Alice and Bob not Figures 1a and 1b. In either agent’s case, the 44
45 (e.g., Conway & Kochen 2006; Fuchs 2010; only perceive, but also act; eigenforms are space X of experiences contains all of the 45
46 Fields 2013); even infants associate agency fixed points of and hence encode regulari- information on which its choices of actions, 46
47 with behavioral unpredictability (e.g., Luo ties in the perception-action relationship. which are assumed to be autonomous and 47
48 & Baillargeon 2010; Csibra & Gergely 2012). Why should such regularities exist? From hence “free,” may be based, including any 48
49 Any black box can, on this view, be consid- the theorist’s point of view, eigenforms memories, values, goals, or other introspec- 49
50 ered to be or at least contain an agent. The are inevitable, as shown by von Foerster tively accessible content. It is important to 50
51 inability of any observer of a black box to (1976) and made more explicit by Kauff- emphasize that a CA does not experience 51
52 determine where in the box an enclosed man (2003, 2009). Such a proof does not, the operations P, D or A, but only the ele- 52
53 agent is, or how much of the box the en- however, say which eigenforms are inevita- ments of the experience space X; an account 53
54 closed agent occupies is what allows the lim- ble. From an agent’s perspective, an eigen- of how experiences are “written on” X is dis- 54
55 iting case in which the other agent is the box form is an eigenpercept, a percept that does cussed below. 55
column A column B column C

Constructivist Foundations vol. 12, N°3

 Second-Order Cybernetics
Eigenforms, Interfaces and Holographic Encoding Chris Fields et al.

column A column B column C

1 « 10 » The analog in Figure 1c of the system affect the state of the other; it thus that human perception is at least approxi- 1
2 arbitrarily-movable inter-agent boundary S “translates” Alice’s actions into Bob’s percep- mately veridical (e.g., Marr 1982; Palmer 2
3 in Figure 1b is the purely notional point at tions and vice-versa, just as the boundary S 1999; Geisler & Diehl 2003; Trivers 2011; 3
4 which Alice’s action A becomes Bob’s per- does in Figure 1b. It plays the role that von Pizlo, Li, Sawada & Steinman 2014) with the 4
5 ception P and vice versa. Consistent with Foerster (1979) assigns, in a very general claim that human perception and action are 5
6 the discussion above, this point is invisible. sense, to language. Either agent’s interac- interactions with a “user interface” formed 6
7 From Bob’s perspective, Alice acts directly tions with its own surface of the blanket can of conscious experience that systemati- 7
8 on his experience space XBob; similarly for be described in terms of Bayesian “active cally hides both the ontology and the causal 8
9 Alice. We can, therefore, simply identify the inference,” in which the agent can choose, structure of the world. As stable action-per- 9
10 two oriented surfaces of the boundary S, given any percept, either to alter its expecta- ception associations, eigenforms “live on” 10
11 the surface facing Bob and the surface fac- tions about the world, i.e., about the prob- this interface. The icons and windows of a 11
12 ing Alice, with the experience spaces XBob abilities of future percepts, or to act in some computer interface are placed there by de- 12
13 and XAlice respectively. In this case, Alice way that changes the percept (Friston 2010; signers. There is, however, no “designer” in 13
14 and Bob each act outwardly, through their 2013). This conceptualization of the agent’s ITP. We discuss in the next section how in- 14
15 own experience spaces, on the experience potential responses to a percept has led to formation can be encoded on an interface by 15
16 space of the other. Note that making this architectural predictions in both neuro- the process of information exchange itself. 16
17 identification of the two surfaces of S with science (Adams, Friston & Bastos 2015) and 17
18 the experience spaces XBob and XAlice renders developmental biology (Friston et al. 2015). 18
19 Alice and Bob neither “open” nor “closed” in « 12 » The idea that perceptions, in the Holographic encoding 19
20 the mereotopological sense (Smith 1996); broad sense of informational inputs from 20
21 Alice and Bob rather share a single bound- the world, appear on a “surface” separating « 14 » Objects as spatially bounded, 21
22 ary that “belongs” to neither of them (for an agent from the world on which it acts – a temporally persistent, internally cohesive, 22
23 further discussion of this point, see Fields surface that not only presents information causally independent entities are simply 23
24 2014). Treating each agent’s outward action and enables action, but also blocks further taken for granted as part of the “classical 24
25 on the other agent as experienced by the epistemic access to what is on the other side worldview” (roughly corresponding to what 25
26 agent performing the action requires giving – immediately suggests a familiar analogy: Edmund Husserl 2012 called the “natural 26
27 the space X a structure that allocates some the user interface of a computer. Like the attitude”) on which human material culture 27
28 part of X for the recording of at least short- surface S in Figure 1b, the user interface of is largely based. This classical conception of 28
29 term memories of executed actions. Record- a computer presents all of the information objecthood is so critical to ordinary human 29
30 ing each action as it is executed, even if this about the computer’s internal state that the cognition that it is widely regarded as innate 30
31 record is “forgotten” immediately thereafter, user can access without disrupting the com- (e.g., Spelke 1994; Baillargeon 2008). Albert 31
32 is the minimal requirement for experienced puter’s function. User interfaces provide Einstein viewed the boundedness, persis- 32
33 learning and hence for experientially under- highly abstracted representations of the tence and causal independence of objects 33
34 standing or expecting anything about the computer’s internal state, each of which al- as critical to science, claiming that “without 34
35 environment. It is, similarly, the minimal re- lows a circumscribed set of possible actions. such an assumption of the mutually inde- 35
36 quirement for any experience of acting, i.e., They systematically hide not just the be- pendent existence (the “being-thus”) of spa- 36
37 of being an agent. havioral complexity, but the entire physical tially distant things, an assumption which 37 203
38 « 11 » The idea that interacting agents and causal structure of the computer. User originates in everyday thought, physical 38
39 interact via a shared, epistemically impen- interfaces are, moreover, ambiguous about thought in the sense familiar to us would not 39
40 etrable boundary has been formulated inde- this structure by design: as with any virtual be possible” (quoted in Fuchs & Stacey 2016: 40
41 pendently by Friston (2013), who provides machine (Smith & Nair 2005), platform in- 6). Niels Bohr (1928, 1958) emphasized that 41
42 an analog, using Pearl’s (1988) Markov blan- dependence is a major component of a user items of laboratory apparatus must be re- 42
43 ket formalism, of the von Foerster–Kauff- interface’s utility. Computer programs are by garded as classical objects if the notion of an 43
44 man demonstration that eigenforms are no means alone in having these properties; “observational outcome” is to make sense. 44
45 inevitable. A Markov blanket is a collec- as Willard Van Orman Quine (1960; see also Eugene Wigner’s (1962) “friend” paradox 45
46 tion of nodes, such that knowing the state Quine 1970) points out, all human natural nicely illustrates the consequences, within 46
47 of this collection renders the states of two languages have them. If a model-theoretic the classical worldview, of not treating other 47
48 sets of nodes interacting only via the blan- approach to semantics (Tarski 1944) is observers as bounded, persistent objects: 48
49 ket conditionally independent (Figure 1d). adopted, all “languages” of any kind have they not only lose any claim to conscious- 49
50 Pearl (1988) shows that a Markov blanket them. A computer’s user interface, however, ness and hence observerhood, they become 50
51 appears whenever a random dynamical sys- obviously has them, which is what makes it a entangled with the rest of the world and ef- 51
52 tem is factored into parts (see Friston 2013 particularly good analogy. fectively disappear. 52
53 or Friston et al. 2015 for more informal dis- « 13 » The Interface Theory of Percep- « 15 » The assumptions of epistemic 53
54 cussions). The blanket effectively encodes tion (Hoffman, Singh & Prakash 2015) transparency and objective persistence over 54
55 information about how the actions of one challenges the still-dominant assumption time underlying the classical worldview 55
column A column B column C

http://constructivist.info/12/3/200.fields
 column A column B column C
1 (a) (b) (c) “does all the work” of observation; the hu- 1
2 man observers read their observational 2
3
Obs Obs Obs outcomes off from the environment in the 3
4 same way that they would read them out of 4
5 a shared or multiply copied book. While the 5
Env Env Env
6 indirectly observed “system” is quantum, 6
7 the directly observed components of the 7
8 Sys Sys Sys environment constitute, in this case, an ef- 8
9 fectively classical object that stands between 9
10 the observer and the quantum system of in- 10
11 Figure 2 • Three views of decoherence. (a) An observer (Obs) prepares and measures a quan- terest. 11
12 tum system (Sys). Both independently interact with a large surrounding environment (Env), « 17 » The environment-as-witness for- 12
13 which renders their states effectively classical by a decoherence mechanism (e.g., ambient mulation of decoherence assumes that the 13
14 photon scattering). (b) The “environment as witness” formulation of Ollivier, Poulin & Zurek observer knows and can characterize the 14
15 (2004, 2005), in which the observer interacts with the system only via the “witnessing” envi- system-environment boundary; the inter- 15
Physics Concepts in Second-Order Cybernetics

16 ronment. This environment decoheres the system but interacts effectively classically with the vening environment is, in other words, as- 16
17 observer. (c) If the assumption of environmental transparency is rejected, the environment sumed to be at least epistemically “transpar- 17
18 becomes a black box. In this case, the system is completely embedded within it in a way that ent.” What happens if this assumption of a 18
19 provides the observer with no access to the system-environment boundary. In this case, deco- transparent environment is rejected? In this 19
20 herence can only be defined at the observer-environment boundary. case, the environment becomes a black box. 20
21 Any “systems” are contained fully within it, 21
22 in such a way that their boundaries, if they 22
23 have them, are observationally inaccessible 23
24 have been criticized at least since Heraclitus. vironment” such as a macroscopic appara- (Figure 2c; cf. Fields 2016a: Figure 1). From 24
25 Quantum theory, however, forcefully raises tus or the ambient photon field interacts the observer’s perspective, it is completely 25
26 the question of how it could even be possible continuously with both the observer and consistent with all available observational 26
27 to experience spatially bounded, temporally the system being observed (Figure 2a; cf. outcomes to treat the “system” as expanding 27
28 persistent, internally cohesive, causally in- Tegmark 2012: Figure 2). This interaction to fill the entire “environment” (formally, 28
29 dependent entities. While some physicists effectively removes quantum coherence system and environment are in an entangled 29
30 still reject it (e.g., Ghirardi, Rimini & Weber from both observer and system by spread- quantum state and so cannot be assigned 30
31 1986; Penrose 1996; Weinberg 2012), unitary ing it over the many unobserved – and in quantum states individually); this is precise- 31
32 quantum theory with no scale-dependent practice unobservable – states of the envi- ly the limiting case discussed above. If the 32
33 physical “collapse” mechanism is increas- ronment (formally, the degrees of freedom system-environment boundary cannot be 33
34 ingly supported by both experiments (e.g., of the environment are traced over). With defined, however, a decoherence interaction 34
35 Eibenberger et al. 2013; Hensen et al. 2015; both observer and observed system now in between system and environment cannot 35
36 Manning, Khakimov, Dall & Truscott 2015; effectively classical states (formally, eigen- be defined either (Fields 2012). Decoher- 36
204 37 Rubino et al. 2017) and theoretical consid- states of their respective interaction Ham- ence can, in this case, only be defined at the 37
38 erations (e.g., Schlosshauer 2006; Tegmark iltonians with the environment), both the observer-environment boundary, i.e., at the 38
39 2012; Saini & Stojkovic 2015; Susskind preparation and measurement interactions interface characterized above. This process 39
40 2016). In unitary quantum theory, the uni- are effectively classical. As pointed out by is illustrated in Figure 3. The quantum state 40
41 verse is permanently in an entangled state; Harold Ollivier, David Poulin and Wojciech Ψ “passes through” the interface to produce 41
42 there are no classical objects. While the ap- Zurek (2004, 2005), however, observers typ- an observational outcome xi. This outcome 42
43 pearance of classicality in such a universe is ically interact with systems of interest only is defined at the observer-environment 43
44 given multiple explanations (for overviews, via an apparatus or an ambient field such boundary (formally, it is an eigenvalue of 44
45 see Landsman 2007; Wallace 2008), since as the photon field (Figure 2b; cf. Ollivier, the observer-environment interaction Ham- 45
46 the 1980s most have appealed in some way Poulin & Zurek 2005: Figure 1). This inter- iltonian). If receiving the observational out- 46
47 to a process of decoherence, i.e., an apparent vening environment serves as a “witness” come xi is to have any determinate effect on 47
48 removal of quantum coherence that results that both decoheres the system and encodes the observer, e.g., if it is to be an input to a 48
49 in an apparently classical object in an appar- information about its state (formally, infor- decision process that selects a next action to 49
50 ently classical state (for reviews, see Zurek mation about the eigenstates of the system- perform, then it must be a classical outcome. 50
51 2003; Schlosshauer 2007). environment interaction Hamiltonian) To characterize xi as classical is just to say 51
52 « 16 » Three views of the decoher- in a way that is accessible to the observer that decoherence actually happens; hence it 52
53 ence process are shown in Figure 2. In the – indeed, to multiple independent observ- is to say that the observer-environment in- 53
54 original environment-induced decoherence ers – via an effectively classical interaction. teraction actually occurs from the perspec- 54
55 process of Dieter Zeh (1970, 1973), an “en- In this picture, the witnessing environment tives of both observer and environment. A 55
column A column B column C

Constructivist Foundations vol. 12, N°3

 Second-Order Cybernetics
Eigenforms, Interfaces and Holographic Encoding Chris Fields et al.

column A column B column C

1 classical outcome can be recorded as a clas- Moore’s (1956) theorem prevents them from 1
2 sical bit string, e.g., a finite sequence of bi- determining any more than a lower limit on 2
3 nary numbers; indeed it must be recorded the number of bulk degrees of freedom of a Quantum 3
4 in a thermodynamically irreversible way black box. The amount of information that state ψ Observer’s 4
xi
5 if it is to be considered to have a causal ef- can be obtained from a black box is strictly outcome xi 5
6 fect (Landauer 1961, 1999; Bennett 2003). limited by the total coding capacity of its 6
7 Where is it encoded? In the CA model, it is boundary degrees of freedom. This coding 7
8 encoded on the space X of experiences. As capacity can be expressed precisely as an ab- 8
9 discussed above, this space X can simply be stract dimension. Let {ξi} be the set of mu- 9
10 identified with the interface. Hence, we can tually independent degrees of freedom of 10
11 regard the classical observational outcome the boundary, and let ni be the number of Interface 11
12 value xi as encoded on the interface itself, as possible distinct values of the i-th boundary 12
13 shown in Figure 3. degree of freedom ξi. The dimension of the Figure 3 • Decoherence encodes a classical 13
14 « 18 » Encodings of classical informa- boundary is then the sum of the numbers ni outcome value xi on the observer-environ- 14
15 tion – information that can be written as a over all the degrees of freedom in {ξi}: ment interface. Such an encoding is required 15
16 finite bit string – on surfaces at which inter- if receipt of the observational outcome is 16
d boundary = / {ξ i} n i
17 actions are defined are called holographic by to be considered to have any effect on the 17
18 physicists. Such holographic encodings were Similarly, let {zj} be the set of mutually-in- observer’s subsequent behavior. This encod- 18
19 first characterized for the surfaces bounded dependent degrees of freedom of the bulk, ing is holographic, i.e., the only information 19
20 by the event horizons of black holes (Beken- and let mj be the number of possible distinct about the environment that the observer 20
21 stein 1973) and were extended to the surface values of the j-th bulk degree of freedom zj. can obtain is the information that can be 21
22 of the observable universe as a whole by Ge- The dimension of the bulk is then: encoded on the observer-environment inter- 22
23 rard ‘t Hooft (1993) and Leonard Susskind face by decoherence. 23
d bulk = / {z j} m j
24 (1995). Holographic encodings record on 24
25 the surface of a system all of the information The amount of information that an observer 25
26 that may be obtained from it by observa- can obtain from any black box is clearly pro- 26
27 tion; to say that a system has a holographic portional to dboundary, not to dbulk; hence any boundary – the interface as described by 27
28 encoding (i.e., satisfies the “holographic black box satisfies the holographic principle. ITP – and shows that the connection is im- 28
29 principle”) is to say that its observationally « 19 » The only information that an plemented by decoherence, the process that 29
30 accessible information content is propor- observer can obtain about the surrounding creates stable classical records of transient 30
31 tional to its surface area, not to its volume environment is the information that can be quantum states. 31
32 (reviewed by Bousso 2002). While the terms encoded on the observer-environment in- 32
33 “surface area” and “volume” here suggest terface by decoherence; the environment of 33
34 ordinary three-dimensional space, the con- any observer is, therefore, a black box and Interfaces encode fitness 34
35 cept of holography is much more general, satisfies the holographic principle (cf. Fields 35
36 applying to any system with a bounding 2016a). The loop from Figure 3 back to Fig- « 21 » As discussed above, information 36
37 surface and an “interior” or as physicists ure 1b is thus closed: from the environment’s is classical to the extent that it has an effect 37 205
38 call it, a “bulk,” that is contained within the perspective, the observer also satisfies the on decision and action, i.e., to the extent that 38
39 boundary. A classical black box provides a holographic principle, as the environment it is useful to the agent that receives it. In- 39
40 suitably abstract example. The boundary can only obtain information about the ob- formation that has no effect – information 40
41 of the black box can be taken to comprise server that can be encoded on the observer- that changes nothing about its recipient – is 41
42 only the degrees of freedom that encode the environment boundary. information that has not been recorded. As 42
43 inputs to and outputs from the box; this re- « 20 » Kauffman (2003, 2011) has previ- Gregory Bateson put it, “what we mean by 43
44 stricted notion of a boundary corresponds ously related the eigenvectors representing information – the elementary unit of infor- 44
45 to the restricted notions of a “system” and an observable degrees of freedom and eigen- mation – is a difference which makes a differ- 45
46 “observer” commonly employed in discus- values representing observable outcome ence” (Bateson 1987: 460; emphasis in origi- 46
47 sions of environment-induced decoherence values in quantum theory to eigenforms nal). All the information that agents possess 47
48 (e.g., Tegmark 2012). In this case, the “bulk” as stable outcomes of repeated measure- is information that has had some effect on 48
49 of the black box comprises all of the non- ments. Indeed, the stability of observational them; it is all “pragmatic information” in 49
50 boundary degrees of freedom, in particular, outcomes under exactly repeated measure- Juan Roederer’s (2005) sense, information 50
51 all of the degrees of freedom involved in the ments underlies the notion of “system prep- that enables doing something. Von Foerster 51
52 process of generating the next output in re- aration” and is often regarded as an axiom of (1970) makes a similar point, quoting Jerzy 52
53 sponse to a given input. It is precisely these quantum theory (e.g., Zurek 2003: 747). The Konorski: “information and its utilization 53
54 “bulk” degrees of freedom to which observ- above discussion localizes this conceptual are inseparable […] one single process” (Fo- 54
55 ers of a black box have no access; indeed connection to the observer-environment erster 1970: 46). 55
column A column B column C

http://constructivist.info/12/3/200.fields
 column A column B column C
1 direction: for every triple of states (x, g, w) independent manner, but show that agents 1
2
W that has occurred so far, F(x, g, w) > 0. It has that make decisions based on these “true” 2
3 not, in particular, allowed an action after world states are generally driven to extinc- 3
4 which no perception follows, or a percep- tion by agents that make decisions solely on 4
P A
5 tion from which no action follows. This the basis of expected fitness (Mark, Marion 5
6 can be expressed probabilistically: an inter- & Hoffman 2010). These empirical results 6
7 face encodes, by its very existence, the fact have since been put on a rigorous footing by 7
8 X G that the probabilities of lethal perceptions a “fitness beats truth” theorem demonstrat- 8
9 D and actions have (at least so far) been low ing that decision strategies based on expect- 9
10 enough that none has occurred. The prob- ed fitness will dominate decision strategies 10
11 Figure 4 • A CA as defined by Hoffman & abilities of perceptions and actions are, how- based on the “truth” about the world for 11
12 Prakash (2014) as a perceive-decide-act (P- ever, specified by the kernels P, D and A and all but a generically small subset of fitness 12
13 D-A) loop through a “world” W, which takes the initial state (x0, g0, w0). If we identify the functions. The “fitness beats truth” theorem 13
14 the place of the “second agent” X2-D2-G2 in interface with X as discussed above, a state x provides a formal justification for von Gla- 14
15 Figure 1c. of X can be viewed as specifying a probabil- sersfeld’s remark that “we must never say 15
Physics Concepts in Second-Order Cybernetics

16 ity distribution Prob(g´| x, g) = D(x, g; g´) of that our knowledge is ‘true’ in the sense that 16
17 the next state g´ of G given the current state it reflects an ontologically real world” (Gla- 17
18 via the Markov kernel D and a probability sersfeld 1981: 93). 18
19 distribution Prob(w´| g, w) = A(g, w; w´) of « 26 » Making use of the computer 19
20 « 22 » In the CA model of Hoffman and the next state w´ of W via the kernel A. Here interface analogy, Hoffman, Singh and 20
Prakash (2014), the recursive loop is per-
21 the kernel action D(x, g; g´) is the probabil- Prakash (2015) characterize perceived “ob- 21
22 ceive-decide-act (P-D-A) as shown in Fig- ity of deciding on g´, given that the current jects” as “icons” on an agent’s interface. 22
23 ure 4. Here perceptions (P) come from and percept is x and the previous decision was g; These icons encode “packages” of expected 23
24 actions (A) are on the “world” W of the CA; similarly for A(g, w; w´). From these an ex- fitness consequences, what James Gibson 24
25 W replaces the “second agent” X2-D2-G2 in pected fitness EF(x | g, w) can be calculated (1979) called “affordances,” though Gibson 25
26 Figure 1c. A CA is defined by the continued by summing over the fitness values of the tended to view affordances as “objectively” 26
27 performance of this P-D-A loop. Should the future states (x, g´, w´) that can immediately encoded by the environment. An icon that 27
28 recursion be for any reason interrupted – follow the current state (x, g, w), with each is a perceived coffee cup, for example, en- 28
29 should there occur a perception after which future state weighted by its probability: codes the expected fitness of its own use for 29
30 no decision follows, a decision after which EF (x ; g, w) = / g lwl F (x, g l , w l ) drinking coffee. They are useful to the extent 30
31 no action (including the action: take no ac- that they support behaviors – at least ap- 31
Prob (g l ; x, g) Prob (w l ; g, w)
32 tion) follows, or an action after which no proximate eigenbehaviors – that leave their 32
33 perception follows – the CA ceases to exist. or making the operator actions explicit: structure at least approximately constant. As 33
34 It is “dead.” EF (x ; g, w) = / g lwl F (x, g l , w l ) noted earlier with respect to experiences of 34
35 « 23 » We can, therefore, define the fit- actions, stable icons representing “objects” 35
D (x, g; g l ) A (g, w; w l )
36 ness of a CA as the probability of continued with “identity over time” or “processes” that 36
206 37 recursion, and the fitness function F of a CA Interfaces, therefore, encode expected fitness. “unfold in time” require some components 37
38 as a mapping F: X × G × W → Non-negative They encode their own best estimates of of the experience set X to be allocated to dis- 38
39 Reals. “Continued recursion” is “viability” their likelihood of survival, i.e., their likeli- tinct collections of “memory” and “expec- 39
40 in Ernst von Glasersfeld’s (1981) sense for a hood of receiving a next input and transmit- tation” experiences. As limits of an infinite 40
41 CA; the CA only survives as long as its P-D- ting a next action. recursive process, as well as fixed points for 41
42 A loop “keeps working.” The meaning of F « 25 » If interfaces encode information that very process, eigenforms are encod- 42
43 becomes particularly clear when the world about fitness, then they do not encode in- ings of their own fitness (F → ∞ in the t → ∞ 43
44 W is regarded as a second agent as in Fig- formation about the observer-independent limit) that the icons manipulated by finite 44
45 ure 1c. The state w of W being such that, for ontology or causal structure of the world. organisms only approximate. 45
46 states x of X and g of G, F(x, g, w) = 0 means In the present conceptual framework, « 27 » It is important to note that the in- 46
47 that the world acts on the agent in a such a of course, this is tautologous: there is no formation about expected fitness that icons 47
48 way that the agent cannot respond. This is observer-independent ontology or causal encode is non-local. Actions taken with re- 48
49 a lethal action. As W is itself defined rela- structure in any world that is defined only spect to one icon can have consequences for 49
50 tive to the agent – it is that agent’s world – W relative to an observer. From the perspec- future interactions with others; one’s actions 50
51 “dies” as well, following such an action. tive of the classical worldview, however, with respect to a perceived kitchen knife, for 51
52 « 24 » We are now in a position to see this is a surprising result. It is supported by example, can have consequences for how 52
53 what interfaces encode. An interface en- evolutionary game-theory experiments that one interacts later with a perceived com- 53
54 codes, by its very existence, the fact that it adopt the classical worldview in so far as puter. An agent that stops interacting, more- 54
55 has not permitted a lethal action in either they assign “true” world states in an agent- over, stops interacting with everything. Such 55
column A column B column C

Constructivist Foundations vol. 12, N°3

 Second-Order Cybernetics
Eigenforms, Interfaces and Holographic Encoding Chris Fields et al.

column A column B column C

1 non-local effects suggest apparent causal This added dimension allows redundancy, 1
2
3
relations between the icons themselves.
Causation in turn suggests an apparent spa-
as shown in Figure 5. An icon that is al-
lowed to occupy space can have “parts” that
4 4 4 4 4 2
3
4 cetime in which causal processes operate. each contribute to the icon’s ability to com- 4
5 Experienced spacetime, however, must be municate a message to the observer. Figure 5 • Spatially encoding an icon allows 5
6 encoded, like the icons themselves, on the « 31 » Redundancy is the key to er- its “parts” to each contribute to its message. 6
7 interface. How is this done? ror correction, and hence to increasing the 7
8 probability that the messages about fitness 8
9 encoded by, for example, “4” and “27” can 9
10 Spacetime as an error- be distinguished. Merely repeating a sym- 10
11 correcting code bol provides the simplest form of redun- dundancy. This added redundancy comes, 11
12 dancy; for example, the code “11” reinforc- however, at a cost: redundant encodings 12
13 « 28 » As noted earlier, the agent-envi- es the message “1.” Three repeats have long require more degrees of freedom and hence 13
14 ronment interface can be characterized in been known to be better than two, as in the a higher dboundary. Distinguishing the values 14
15 abstraction from any notion of ordinary long-standing Morse-code emergency dis- of these additional degrees of freedom re- 15
16 three-dimensional space. Human percep- tress signal: quires, moreover, an energy expenditure of 16
17 tion, however, is resolutely spatial: the at least N × ln2 × kT per distinction, where 17
···–––···
18 “objects” we see occupy space and move in N is the number of bits required to encode 18
19 space, and the actions we take are taken in or “SOS,” by convention always repeated each distinguishable value, k is Boltzmann’s 19
20 space. Human experience, moreover, un- three times. constant and T is absolute temperature 20
21 folds in time. Where does this spacetime « 32 » To examine the use of redundan- (Landauer 1961, 1999; Bennett 2003). 21
22 come from? The recursion that gives rise to cy, we first consider the simplest case, a bi- « 35 » Organisms such as humans do 22
23 eigenforms provides a natural “counter” for nary code. For a binary code, the Hamming not encode one-to-one eigenform-to-ei- 23
24 time; this conception of time as an agent- distance provides a convenient measure of genbehavior relationships: there are many 24
25 specific counter for experience is built into the dissimilarity or distance between two different uses for a screwdriver or a coffee 25
26 the CA framework (Hoffman & Prakash encoded symbols. The codes “111” for “S” cup, and one can reach for and grasp many 26
27 2014). What, however, about space? What and “000” for “O” are, for example, sepa- different objects. We suggest that organ- 27
28 is it about perception-action interfaces rated by a Hamming distance of three; three isms faced with the task of encoding such 28
29 that makes them spatial, and what explains bit flips are required to transform one mes- complex relationships devote some of their 29
30 three-dimensionality? sage into the other. The redundancy of such available interface redundancy to encod- 30
31 « 29 » We suggest that space, and by a code provides a natural sense of spatial ing eigenform persistence over time and 31
32 extension spacetime, provides an error- dimensionality, as shown in Figure 6. Here the rest to encoding eigenform actionabil- 32
33 correcting code for fitness consequences. A flipping a bit is “traveling” in a “direction” ity. For example, some degrees of freedom 33
34 spatiotemporal encoding provides a way of on a graph. The bits are independent, so the are devoted to encoding that a coffee cup 34
35 “spreading out” information about fitness directions are orthogonal. is present, while others are devoted to en- 35
36 in a way that allows redundancy and hence « 33 » As can be seen in Figure 6, a coding whether and how it can be grasped. 36
37 an ability to detect and correct perceptual three-bit binary code provides the possibil- Encodings of persistence and actionability 37 207
38 errors. To see the value of a spatial encod- ity of error correction – every message with are subject to different constraints. An ac- 38
39 ing, consider the information about quan- mixed bits has a 67% likelihood of being tion type, like grasping, may be executed in 39
40 tity encoded by the positive whole numbers. one pure-bit message and only a 33% likeli- a large number of ways, only one of which 40
41 These numbers are just discrete points on hood of being the other – while the two-bit may yield positive fitness (getting one’s cof- 41
42 the real line, hence they can be represented code does not. Hence a three-fold redun- fee!) in a particular situation. Accurately se- 42
43 simply as a sequence of points: dancy is the minimum for error-correction lecting the one right high-fitness grasp from 43
44 utility for a binary code. the large number of possible grasps requires 44
··················
45 « 34 » At the very basis of human per- a redundant encoding, but redundantly en- 45
46 This representation can even be compressed ception is a binary question: is something coding many distinct grasps is expensive. 46
47 further: there or not? It is this question that distin- One might expect, therefore, organisms to 47
48 guishes an “object” from an undifferentiated employ the minimal redundancy that pro- 48
·
49 “background.” We suggest that the need to vides error correction, three-fold redun- 49
50 « 30 » Such representations are, how- answer this simple binary question accu- dancy, for action encoding. Assuming a 50
51 ever, useless: there is no way to tell, for ex- rately requires the error-correction capabil- continuous range of grasps, a three-fold re- 51
52 ample, that “·” represents 4 while “·” repre- ity of a triply redundant encoding and hence dundant encoding is an encoding into real 52
53 sents 27. Making this distinction requires a three-dimensional Hamming space. Sys- ordered triples and hence into real three- 53
54 adding a spatial dimension that allows a tems that must answer more complex ques- space. Discretizing the possible grasps vox- 54
55 planar character like “4” to be drawn out. tions can be expected to employ greater re- elates this space. 55
column A column B column C

http://constructivist.info/12/3/200.fields
 column A column B column C
1 « 36 » Employing a distinct real or « 38 » Mammalian visual (e.g., Goodale information about fitness and hence per- 1
2 even a high-resolution discrete three-space & Milner 1992) and auditory (e.g., Hickok & sistence. Spacetime itself, therefore, is an 2
3 for each of a large number of action types Poeppel 2007) systems use distinct process- encoding of fitness; it exists only because it 3
4 would, however, be very expensive both for ing streams for action and object perception, is useful to organisms going about the busi- 4
5 encoding perception and for memory; one consistent with the prediction above. Ob- ness of staying alive. Organisms with differ- 5
6 would therefore expect organisms to overlay jects are indeed categorized quasi-hierarchi- ent structures and lifestyles – as different as 6
7 their encodings so as to encode many differ- cally (e.g., Martin 2007). The shapes of both E. coli, an oak tree, and a person – may expe- 7
8 ent action types in the same space. Whether natural and artificial objects can often be rience very different “spacetimes.” 8
9 this is possible depends on the composabil- represented by scalable codes such as crys- « 41 » It remains, however, to extract 9
10 ity of actions and the existence of inverse ac- tal structures, Fibonacci numbers or fractals from this idea predictions of sufficient 10
11 tions, i.e., on whether the action space sup- (e.g., Thompson 1945; Mandelbrot 1982). power and precision that confirming them 11
12 ports a group structure. It has been shown, The idea that spacetime itself is emergent would overcome the intuitive appeal of an 12
13 within the CA framework, that a group from underlying quantum- or information- “objective” spacetime filled with “objec- 13
14 structure on the action space G induces one theoretic constraints is now being taken tive” objects. The stubborn resistance of the 14
15 on the interface X (Hoffman, Singh & Pra- seriously by physicists (e.g., Swingle 2012; classical worldview in the face of eight de- 15
Physics Concepts in Second-Order Cybernetics

16 kash 2015). Hence it is plausible to suggest Arkani-Hamed & Trnka 2014; Pastawski et cades of quantum theory, experiments and 16
17 that three-fold encoding redundancy and al. 2015; D’Ariano & Perinotti 2017). technology shows that this will not be easy. 17
18 a group structure on actions is sufficient to Bringing these ideas into the science – and 18
19 generate an interface with three extended hence the technology – of perception itself 19
20 “spatial” dimensions in which actions are Conclusion may yet, however, open the door to empiri- 20
21 represented. cal demonstrations that cannot be denied. 21
22 « 37 » The encoding of eigenform per- « 39 » In his paper introducing the “it 22
23 sistence, on the other hand, is subject only from bit” concept, John Wheeler insisted 23
24 to the constraint of being “good enough” to that “what we call existence is an informa- Acknowledgements 24
25 support appropriate actions. One can, there- tion-theoretic entity” (Wheeler 1990: 8), 25
26 fore, expect a quasi-hierarchical encoding in later quoting Gottfried Leibniz, “time and We thank Federico Faggin and Manish 26
27 which resolution can be varied to suit obser- space are not things, but orders of things” Singh for discussions and the three anony- 27
28 vational context. As this encoding must “fit and Einstein, “time and space are modes mous reviewers for their comments. C. F., D. 28
29 into” a spatially-organized interface, one ex- by which we think, and not conditions in D. H. and C. P. thank the Federico and Elvia 29
30 pects a spatial encoding in which the spatial which we live” in support of his “Fourth No: Faggin Foundation for financial support. 30
31 dimensions associated with a particular ei- no space, no time” (ibid: 10). Von Foerster 31
32 genform are not extended over the entire in- could well have added: spacetime is the ei- Received: 20 November 2016 32
33 terface but are rather “compressed” into only genform that by remaining constant enables Accepted: 7 April 2017 33
34 a small part of the interface. A compressed actions. 34
35 spatial structure is a shape, like “4” in Fig- « 40 » To this we have added: eigen- 35
36 ure 5, that occupies space and redundantly form – eigenbehavior loops, and hence the 36
208 37 encodes persistence. interfaces through which they pass, encode 37
38 38
39 39
40 40
41 41
42 42
43 43
44 44
45 45
46 46
47 47
48 48
49 49
50 50
51 51
52 52
53 53
54 54
55 55
column A column B column C

Constructivist Foundations vol. 12, N°3

 Second-Order Cybernetics
Eigenforms, Interfaces and Holographic Encoding Chris Fields et al.

column A column B column C

{
1 1
2 Chris Fields 2
3 PhD Philosophy, University of Colorado 1985, is an information 3
4 scientist interested in the physics, developmental biology and 4
5 cognitive neuroscience of object perception and object re- 5
6 identification over time. His recent publications include work in 6
7 the foundations of quantum theory, endophysics, morphogenesis, 7
8 cognitive modeling, and the etiology of autism spectrum disorders. 8
9 9
10 10

}
11 11
12 Donald D. Hoffman 12
13 PhD Computational Psychology, MIT 1983, is a cognitive scientist and author of more than 13
14 100 scientific papers and three books, including Visual Intelligence: How We Create What 14
15 We See (2000). He joined the faculty of UC Irvine in 1983, where he is now a full professor 15
16 in the departments of cognitive science, computer science and philosophy. He received a 16
17 Distinguished Scientific Award of the American Psychological Association for early career 17
18 research into visual perception, the Rustum Roy Award of the Chopra Foundation, and 18
19 the Troland Research Award of the US National Academy of Sciences. Hoffman’s research 19
20 has led to a “user interface” theory of perception, which proposes that natural selection 20
21 shapes our perceptions not to report truth but simply to guide adaptive behavior. 21
22 22
23 23

{
24 24
25 Chetan Prakash 25
26 PhD Mathematical Physics, Cornell University 1982, has published, with Bruce 26
27 Bennett and Donald Hoffman, the book Observer Theory. His current research intends 27
28 to elaborate a theory that shows how consciousness gives rise to the “physical” 28
29 world as our interface with reality – as against the idea that brains produce 29
30 consciousness. As this “reverse hard problem of consciousness” is a view by no 30
31 means standard in the scientific community, he has used rigorous mathematical 31
32 analyses to demonstrate the falsity of the commonly held belief that evolution 32
33 has led us to perceive an “objective” reality with ever-increasing accuracy. 33
34 34
35 35

}
36 36
37 Robert Prentner 37 209
38 PhD Physical Chemistry, ETH Zürich 2013, has been a visiting scholar 38
39 at Stanford University’s Center for the Explanation of Consciousness. 39
40 Since Fall 2013 he has been working at the Department of 40
41 Humanities, Social and Political Sciences at ETH Zürich continuing 41
42 his philosophical studies and lecturing in the philosophy of science. 42
43 He is member of the editorial office of the journal Mind and Matter. 43
44 44
45 45
46 46
47 47
48 48
49 49
50 50
51 51
52 52
53 53
54 54
55 55
column A column B column C

http://constructivist.info/12/3/200.fields
 column A column B column C

Open Peer Commentaries

1 1
2 2
3 3
4 4

on Chris Fields et al.’s “Eigenforms, Interfaces and Holographic

5 5
6 6
7 7
8
9
Encoding” 8
9
10 10
11 11
12 12
13 13
14 14
15 15
Physics Concepts in Second-Order Cybernetics

16 16
17 Do Nonclassical Worlds taken by paraconsistent logicians, especially experiments from their theory. For all they 17
those who embrace dialetheism, the thesis know now, it seems, the classical world is the
18
19
Entail Dualism? that some contradictions are true, while also world, or at least one of them. There are not
18
19
20 Eric Dietrich being false. So, for example, to a dialetheist, merely different viewpoints, rather there are 20
21 the Liar Paradox – “This sentence is false” different worlds. 21
Binghamton University, USA
22 – is both true and false at the same time. In « 4 » The authors, then, are stuck with 22
23
dietrich/at/binghamton.edu this commentary, I argue that though the the classical, mind-independent world while 23
24 authors opt for an eliminativist approach they develop and experimentally test their 24
25 > Upshot • The vast differences between to the nonclassical-classical paradox, they new theory, which posits a nonclassical, 25
26 the objective, classical realm of our ev- ought to opt for the last way: they ought to mind-dependent “world” as a replacement. 26
27 eryday lives and any nonclassical realm embrace the dualistic paradox. « 5 » It is not clear what the authors 27
28 (like quantum physics) have worried « 2 » In their article, Fields et al. present hope for at this stage. They themselves are 28
29 researchers for almost a century. No at- an interesting and large theory that begins acutely sensitive to the staying power, the 29
30 tempt at resolving the differences or with taking observer-relativity seriously and stubbornness, of the classical world. But 30
31 explaining them away has ever worked. ends with the proposal that spacetime could they also know the explanatory power of 31
32 Maybe there are two realms, the classical profitably be construed as error-correcting mind-dependent approaches to under- 32
33 and the nonclassical, and maybe they are code. Then at the end, in §41, the authors say standing minds and their realities (there 33
34 paradoxical. that their theory still needs to produce pre- are many reasons to take observer-relativity 34
35 dictions sufficiently powerful to overcome seriously). One gets the impression that by 35
36 « 1 » Chris Fields et al. are wrestling the intuitive appeal of mind-independent drawing from several sources – quantum 36
210 37 with, among other things, the paradox, the spacetime filled with mind-independent physics, consciousness studies, cognitive 37
38 clash, between “quantum reality” and “clas- objects – i.e., powerful enough to overcome science, evolutionary theory, math, and phi- 38
39 sical reality” concerning tables and chairs our resolutely perceiving the classical world. losophy – the authors hope that their theory 39
40 and dogs and cats and people. There are « 3 » In the very next sentence, the will simply liberate the human mind from 40
41 usually two main ways to deal with paradox. reader senses perhaps some despair on the its preference for occupying a mind-inde- 41
42 One can try to explain it away (the para- part of the authors, for they bemoan the pendent universe. 42
43 dox is illusory) or one can try to eliminate “stubborn resistance” of the classical world « 6 » At this point a movie reference is 43
44 it by showing that one side of the paradox in the face of eight decades of quantum needed. In the movie Arrival, space aliens 44
45 is based on a mistake. Optical illusions theory – in effect saying that after eight show up in the present time and offer us the 45
46 are one example of the former way; Zeno’s decades, one would have thought that we gift of their written language. This language 46
47 Paradoxes of Motion are an example of the would have finally said goodbye to the clas- is unlike any language on Earth. To use it, 47
48 latter way. Of course, there are other, less sical world, to the mind-independent world. one has to have a decidedly nonhuman re- 48
49 common ways of dealing with paradox. One Interestingly, perhaps in an effort to hurry lation to time – in particular, one has to be 49
50 can just stipulate the paradox away. This is the classical world out the door, the authors able to see the future. To the space aliens, 50
51 the method used by mathematicians when do not use the term “world,” but rather call seeing the future is second nature; indeed, 51
52 dealing with the paradoxes of set theory; it a worldview. But this latter is a term they they experience all at once what we would 52
53 this method really only works if one is pre- are not entitled to because, as they just said, call sequential events. The key is that when 53
54 pared to go axiomatic. And lastly, one can they have yet to prove their theory experi- humans learn the alien language, their 54
55 just embrace the paradox. This is the way mentally because they have yet to derive any perception of time changes, and, like the 55
column A column B column C

Constructivist Foundations vol. 12, N°3

 Second-Order Cybernetics
Do Nonclassical Worlds Entail Dualism? Eric Dietrich

column A column B column C

1 space aliens, they then also see the future, Everything in the video is classical. The tradictions). As with the other issues in this 1
2 experiencing all events at once. Learning Higgs is not. The same can be said of X. area, it is not clear why this meta-world ap- 2
3 their language changes our brains. Do the « 9 » So, what to do? We humans seem pears or exists. I am inclined to invoke the 3
4 authors want the same property for their to occupy one realm, the classical one, observer, which is what I think the authors 4
5 theory – merely learning it, or learning that while developing nonclassical theories of might support.) 5
6 experiments support it, will change our nonclassical realms accessible to us only via « 13 » The cost associated with this 6
7 stubborn human resistance to sensing the our thought (the LHC is classical, the data contradictory-worlds approach, and not 7
8 world in a mind-dependent fashion? Will from its experiments are classically pre- just contradictory points of view, is that 8
9 learning their theory, or learning that their sented and represented, but via our minds, consciousness remains unexplainable. But 9
10 theory agrees with all experimental chal- we see beyond the data to a nonclassical many of us already think this is the ulti- 10
11 lenges, change human perception in such a world). And the two realms together form mate knowledge about consciousness (Di- 11
12 way that the classical world is eliminated? a paradox: crucial propositions true in one etrich & Hardcastle 2005). It is unlikely 12
13 « 7 » Of course, it is unlikely the au- realm are false in the other. that the authors will agree with this since a 13
14 thors want any such thing (still, in §41, « 10 » One proposal is to give up the large part of the motivation for their article 14
15 they do say that confirmatory predictions quest to “overcome the intuitive appeal” of is bringing consciousness into the science 15
16 of their theory would “overcome the in- the classical world (§41). Embrace the two tent. 16
17 tuitive appeal of [the classical world …]”). worlds, or many worlds, solution: one is « 14 » Regardless of whether one picks 17
18 Assuming the authors do not think mere classical and others are not. one world with many contradictory, para- 18
19 knowledge of their theory will liberate hu- « 11 » Specifically, the authors’ theory doxical viewpoints or many contradictory 19
20 mans from our classical world or diminish could explain human and other animal worlds, the (unintended) message of the 20
21 its appeal (this has not worked for quantum minds in the nonclassical way they detail, authors’ research seems clear: the classical 21
22 mechanics), then what are they going to do while at the same time, we humans and world does not merely have an “intuitive 22
23 about “the stubborn resistance of the classi- other cognizers occupy a classical world. appeal” for us (§41), rather it is ineluc- 23
24 cal world”? Unless something frees human « 12 » I said above (in §3) that the au- table. We are classical beings with minds 24
25 perception from its moorings in the clas- thors were not entitled to use the term “clas- that allow us to see the nonclassical. How 25
26 sical world, it does not matter what brave sical worldview” (from their §41) because this can be so is very puzzling. And the 26
27 new theory is developed, the moorings will until their theory was supported by experi- authors’ theory does not directly address 27
28 remain. ments, they could not know that classical- this. However, as already claimed in §8 28
29 « 8 » Suppose X is the extremely so- ity was a worldview and not a world. We above, it is very unlikely that any theory of 29
30 phisticated future version of the theory the now see that “classical worldview” has an- this “dualism” – classical beings studying a 30
31 authors are working on now, and thus con- other problem. It suggests that there is one nonclassical realm – will ever be intuitive 31
32 sidered “ultimately true.” It is profoundly world: from one worldview (point of view) to us even though it may well be robustly 32
33 unlikely that X will finally free humans it looks classical and from another it looks explanatory. What will come to seem in- 33
34 from their classical worldview (the mov- nonclassical. Think about walking around tuitive then is that what is called “reality” is 34
35 ie, after all, was fiction). Rather, we will a car. From one view (a sideview) the car bigger than we thought, and more unstable 35
36 be stuck with the very situation we have looks one way, from another (a front view) and protean than we supposed. Epistemic 36
37 now with quantum physics, where human it looks another. The “real” car is the inte- humility should follow. 37 211
38 physicists occupy the classical world while gration of all such views (for the viewer). 38
39 they develop, experiment on, and prove the Note that the car is not paradoxical, so the Eric Dietrich is professor of philosophy at 39
40 nonclassical theory of quantum reality. We integration works. But this does not apply Binghamton University in New York and the editor 40
41 have had 80 years of quantum mechanics to the world posited by the authors’ and the of the Journal of Experimental and Theoretical 41
42 (as Fields et al. note). In that whole time, no classical view we inhabit as we read about Artificial Intelligence. He is the author of the book 42
43 physicist has started experiencing the non- their theory: the two are decidedly para- Excellent Beauty: The Naturalness of Religion and 43
44 classical world in their daily lives. Rather, doxical. So, integration is unlikely to work. the Unnaturalness of the World (2015) and the paper 44
45 they all daily experience the classical world. The one-world-with-two-worldviews ap- “There is no progress in philosophy” (2011). 45
46 And these physicists also experience the proach might, I suppose, better accord with 46
47 classical world while they experiment on Ockham’s Razor, but that’s not in the cards. Received: 31 May 2017 47
48 and theorize about the nonclassical world. This all suggests that there are many worlds Accepted: 8 June 2017 48
49 So, the authors’ theory, X, will represent a – we view them somehow by visiting them, 49
50 nonclassical realm, and we will learn it, ap- by “changing locations,” via our conscious- 50
51 ply it, and come to see X’s beauty, all the ness. (Of course, ontologically, some of us 51
52 while firmly planted in the classical world. are still committed to some over-arching, 52
53 Go back and watch the videos of the an- single meta-world, and this meta-world has 53
54 nouncement at the Large Hadron Collider to be at least contradictory and probably 54
55 (LHC) of finally finding the Higgs Boson. dialetheic (the locus of unresolvable con- 55
column A column B column C

http://constructivist.info/12/3/200.fields
 column A column B column C
1 Where is Spacetime sciousness constitutes the world. The au- place in an abstract mathematical space 1
thors avoid the presupposition of “objects rather than in the space of our everyday
2
3
Constituted? as spatially-bounded, temporally-persistent, world, the question arises: What is the onto-
2
3
4 Urban Kordeš internally-cohesive, causally-independent logical status of entities or phenomena those 4
5 entities” (§14), and instead attempt to create spaces represent? 5
University of Ljubljana, Slovenia
6 a mathematical model of the constitution of 6
7
urban.kordes/at/pef.uni-lj.si those objects, presuming the primacy of con- The gap between functional 7
8 sciousness. Discarding the natural attitude and phenomenal aspects 8
9 > Upshot • In an attempt to understand (the tendency to believe our construction of of consciousness 9
10 its presuppositions, the commentary the world to be an accurate representation « 8 » Susan Blackmore (2013) divides 10
11 takes a closer look at the model proposed of objective reality), the authors seem to as- discussions concerning consciousness into 11
12 by the target article. By analysing the sume the phenomenological attitude (Hus- two distinct realms represented by the fol- 12
13 interactions between conscious agents, serl 1982), the attitude that phenomenol- lowing two questions: “What is it like to 13
14 the model tries to derive the enaction of ogy shares with constructivism (as argued in be…?” and “What does consciousness do?” 14
15 a spacetime framework. A critical exami- Kordeš 2016a). (for the purposes of this commentary, they 15
Physics Concepts in Second-Order Cybernetics

16 nation of the ontological status of the in- « 4 » According to phenomenology, will be referred to as the phenomenal and 16
17 volved entities indicates inconsistencies, phenomenal consciousness is the episte- the functional aspect respectively). There 17
18 especially at the adoption of viewpoints. mologically safest foundation on which are many answers to the latter. One of them 18
19 It seems that despite the model’s being to build science. According to Dan Zahavi is proposed by the target article, i.e., con- 19
20 supposedly grounded on the primacy of (2004), for Edmund Husserl, studying how sciousness behaves in principle unpredict- 20
21 consciousness, this characteristic is not the world is constituted in consciousness ably. Between the functional aspect of con- 21
22 immediately apparent. The commentary became the cornerstone for transcendental sciousness and the aspect that answers the 22
23 proposes an even more radical adoption phenomenology, which in turn was sup- question “What is it like to be…?” (describ- 23
24 of the first-person point of view. posed to become the foundation of science. ing so-called phenomenal consciousness), 24
25 Despite the fact that Husserl created a philo- there is an unsurmountable chasm – usually 25
26 Ontological status of entities in the sophical system with this particular pur- referred to as the explanatory gap. 26
27 conscious agents model pose, phenomenology has never completely « 9 » In order to assess which aspect is 27
28 « 1 » I am inclined to support the model succeeded in this endeavour. The problem assumed by the authors of the target article, 28
29 presented by Chris Fields et al., especially the being that phenomenologists never made it the basic mathematical elements of the pro- 29
30 way it, in one big stroke, connects biological exactly clear how to actually build natural posed model need to be examined. What are 30
31 constructivism (Maturana & Varela 1980; science (starting with physics) on phenom- the categories that define agents CA1 and 31
32 Foerster 1984; Riegler 2012) with quantum enological foundations. The target article of- CA2, the interaction between whom enacts 32
33 physics. Yet, extraordinary claims (such as fers a solution. physical entities? Figure 1 of the target ar- 33
34 the proposed model) require extraordinary « 5 » The proposed mathematical model ticle provides the answer: “Here X1 and G1 34
35 evidence. When the model’s results confirm is based on the concept of conscious agents and X2 and G2 are measurable spaces rep- 35
36 the authors’ goals, i.e., that from the interac- (CAs) (§2). In the following paragraphs I resenting the experiences and available ac- 36
212 37 tions of conscious agents almost miraculous- will try to summarise and more clearly ex- tions, respectively.” The space X is especially 37
38 ly springs a 3 + 1D physical framework of our plicate the presuppositions that come with important as on it rests the weight of the en- 38
39 everyday world, one should always beware of this concept. tire model. It is precisely X that is supposed 39
40 the possibility of motivated reasoning. « 6 » The authors suggest that a defining to contain encoded objects. 40
41 « 2 » As the remainder of this section feature of a CA is its “principled unpredict- « 10 » But what kind of entities does X 41
42 will show, an explication of the proposed ability […] considered by some to indicate represent? What is the meaning of “expe- 42
43 model’s presuppositions exposes consider- autonomy or ‘free will’ and hence agency riences” (§9) within the model? It would 43
44 able issues. It remains to be seen whether from the perspective of external observers” seem that X also introduces phenomenal 44
45 those problems stem from the commenta- (§6). Furthermore: consciousness into the model based on the 45
46 tor’s misunderstanding, from small incon- strong presupposition that phenomenal 46
47
48
sistencies in the proposed model (which can
be easily patched), or from flaws with serious
“ While autonomy in the non-trivial machine
sense inferred above is somewhat abstract, a re-
consciousness can be mathematically de-
scribed. With this, the model adopts the
47
48
49 consequences for the model’s fitness. I hope quirement for autonomous decision-making at first-person perspective of lived experience 49
50 it will turn out to be one of the former op- least suggests an awareness of potential conse- (a perspective that is unreachable for most 50
51
52
tions, for the idea of deriving characteristics
of the physical world from the dynamic of
”
quences and hence consciousness. (§8) of natural science). By simultaneously in-
cluding the functional and the phenomenal
51
52
53 consciousness is an exceptional one. « 7 » From this definition of a CA, it is aspect of consciousness it seems that the 53
54 « 3 » The aim of the target article is to clear that consciousness is inferred from the model of Fields et al. unwittingly mixes first- 54
55 create a mathematical model of how con- CA’s behaviour. Since this behaviour takes and third-person perspectives. 55
column A column B column C

Constructivist Foundations vol. 12, N°3

 Second-Order Cybernetics
Where is Spacetime Constituted? Urban Kordeš

column A column B column C

1 « 11 » Another indication for the mix-up enables intersubjectively valid methods and world. (One of such notable attempts being 1
2 of perspectives is the model’s separation of G exceptionally successful research, character- “primary algebra” proposed by George Spen- 2
3 and X. Separating experiences (X) and avail- istic of physics, neuroscience, biology, etc. cer Brown 1969 in his Laws of Form). 3
4 able actions (G) indicates a distinction be- What this view filters out, though, is con- « 18 » It would seem that the authors are 4
5 tween the two. If the model took the auton- sciousness. It perceives the researched struc- not modelling the construction of a world 5
6 omy of CAs and the primacy of experience tures as “real” and forgets that they came from consciousness, but the construction 6
7 seriously, G would be a subset of X – available about only due to the act of consciousness. of a world by entities that are behaving as if 7
8 actions are only those noticed or autono- If naturalising research approaches are at all conscious. 8
9 mously constructed and as such experienced interested in consciousness, they look for it 9
10 as available by the CA. Because that is not the as a product of those natural structures. By Agency and the sense of agency 10
11 case, the only possible interpretation is that filtering out the observer’s consciousness, « 19 » The confusion of perspectives is 11
12 the authors presuppose the possibility of a the naturalistic view can only resort to infer- also apparent from the use of the term “agent” 12
13 space of available actions as perceived from ence from behavior when trying to detect and the consequential notion of agency. It 13
14 outside the CA. This takes autonomy away consciousness “out there.” As a consequence, seems that the authors conflate the sense of 14
15 from the agent. Being autonomous means they can only answer the functional question, agency with agency as the actual ability of a 15
16 that the agent chooses from the options the i.e., “What does consciousness do?” while CA to consciously influence courses of ac- 16
17 agent itself constructs rather than from pre- the question of phenomenal consciousness – tion. Agency and the sense of agency should 17
18 given options (cf. Winograd & Flores 1986). “What is it like to be…?” – is inaccessible to not be carelessly equated. Many third-person 18
19 Genuine autonomy is in the very construc- the behaviour-oriented third-person view of studies such as those of Benjamin Libet et 19
20 tion of the elements of the world, which are, natural science. al. (1983) and Daniel Wegner (2003) have 20
21 in this case, options to choose from. « 15 » By renouncing the view from no- shown that our conscious decisions are not 21
22 where, consciousness appears everywhere. (always) causally linked with our actions, de- 22
23 Consciousness as the foundation Phenomenal consciousness imbues every- spite what the sense of agency might suggest. 23
24 « 12 » With the exception of phenom- thing there is, everything one notices, thinks The phenomenal sense of agency functions 24
25 enology, most other approaches see con- or perceives (Kordeš 2016b). Consciousness mostly as a way of smoothing the narrative 25
26 sciousness as a product of an observer- from the first-person perspective is a me- (i.e., sense-making). 26
27 independent, “natural” world (i.e., they dium in which all features of the world are « 20 » Agency and the sense of agency 27
28 naturalise consciousness). If consciousness constituted. could only be equated if the model were to 28
29 is to be taken as the foundation of a theory, « 16 » The history of cognitive science be intrinsically rooted in the experiential 29
30 then naturalising approaches are inappropri- has shown that the growing understanding world, that is, if the whole process were to 30
31 ate, as they presuppose the primacy of some- of brain dynamics and human behaviour be seen as metamorphoses of phenomenal 31
32 thing other than consciousness. The only does not bring us closer to understanding consciousness. Such a model would describe 32
33 aspect of consciousness that can be used as experience. The failure to bridge the explana- a consciousness that changes itself. That way, 33
34 the foundation for a theory is phenomenal tory gap points towards the conclusion that sense-making, the constitution of objects, 34
35 consciousness, i.e., lived experience. This is phenomenal consciousness is not only pri- etc. would all be part of the same substance, 35
36 only possible if the theory’s point of view is a mary but also irreducible. If we want to get and the dualism that spoils the image of the 36
37 first-person one. However, in the case of the conscious experiences as a result, we have presented model would be avoided. 37 213
38 proposed model it is the point of view of the to start with conscious experiences. Only in 38
39 CA. that case can we say that we take conscious- Acknowledgement 39
40 « 13 » Constructivists always stress that ness as the foundation of our theory. I would like to thank Florian Klauser for 40
41 every view is a view from somewhere. I « 17 » The model proposed in the target his help with exploring and articulating the 41
42 fear that Fields et al. are not very clear from article puts agents and their life dramas in an target article’s (as well as our own) epistemo- 42
43 where they are observing. Are they looking abstract space. The authors attempt to “de- logical positions and presuppositions. 43
44 at the world from the eyes of an agent (who, velop the dynamics of interacting conscious 44
45 of course, does not have access to anything agents, and study how the perception of ob- Urban Kordeš is professor of cognitive science and 45
46 other than its own horizon – i.e., the surface jects and spacetime can emerge from such first-person research at the University of Ljubljana where 46
47 that connects it to the world) or through the dynamics” (Hoffman & Prakash 2014: 557). he is currently heading the cognitive science programme. 47
48 “eyes of God,” who sees all agents, their ac- Whatever this space is supposed to represent His current research involves such training in order to 48
49 tions and interactions? does not seem to represent the space of phe- study the phenomenology of the enactment of knowledge. 49
50 « 14 » The “God’s eye” view or the view nomenal consciousness. As argued above, 50
51 “from nowhere” (Nagel 1989) is character- only if the theory performs the (very radi- Received: 6 June 2017 51
52 istic of fields that have uncritically accepted cal) step of grounding itself in phenomenal Accepted: 14 June 2017 52
53 the natural attitude (that is, for most of sci- consciousness, is it sensible to start looking 53
54 ence with a few exceptions, such as phenom- for appropriate mathematics that might en- 54
55 enologically inspired research). This view able the modelling of the constitution of the 55
column A column B column C

http://constructivist.info/12/3/200.fields
 column A column B column C
1 Eigenform Encoding to itself. Alas, this fixed point will not be way with his statement “I am the observed 1
reached if we take a starting value that is not relation between myself and observing my-
2
3
and Spacetime equal to 1. If we start with a number greater self ” (Foerster 1981). We can go from von
2
3
4 Louis H. Kauffman than 1 and square it, we get a number even Foerster to Wheeler by a substitution: “The 4
5 greater than that and the values will ap- Universe is the observed relation between 5
University of Illinois at Chicago, USA
6 proach infinity. Infinity! Well we were not itself and observing itself.” There is no dif- 6
7
kauffman/at/uic.edu thinking of that as a number, but surely In- ference. Spacetime, the Universe, the Self, 7
8 finity2 = Infinity and so Infinity is (if we al- all are central eigenforms in the genesis of 8
9 > Upshot • An eigenform is both a sym- low it into our conversation) an eigenform worlds. These words are here capitalized to 9
10 bol for a process and the instantiation of for T. If we take a number greater than 0 and indicate their roles in this allegory of the na- 10
11 a process itself. As such, eigenform pro- less than 1, then applying T to that number ture of Everything. 11
12 vides a new entry to spacetime, as a uni- will lead to a sequence that tends to 0. And 12
13 fication of entity, place and process. 0 is a fixed point of T, indeed. So, we have Quantum theory and it from qubit 13
14 found that T has three eigenforms, Infin- « 4 » Having stated my point of view, 14
15 What is an eigenform? ity, 1 and 0. This could lead us out beyond directly and allegorically, let us turn to the 15
Physics Concepts in Second-Order Cybernetics

16 « 1 » In order to provide some back- the specific transformation to thoughts target article, where the authors say “[…] we 16
ground for a discussion of the target article
17 about the fantastic distinction that seems to pursue the notion of an eigenform not from 17
18 “Eigenforms, Interfaces and Holographic present itself between the Infinite, the Noth- the point of view of an abstract reflexive 18
19 Encoding” by Chris Fields, Donald Hoff- ingness and Unity. We could go off track as system, but from von Foerster’s original per- 19
20 man, Chetan Prakash and Robert Prentner, I far as the calculating forms are concerned spective of an agent that observes and acts 20
21 shall start this discussion by describing what and find that the simple working with and on its world” (§2). This is a correct stance. 21
22 an eigenform is and then I shall explore the searching for a fixed point for T(x) = x2 has One can consider an abstract reflexive sys- 22
23 nature of the relationship between quantum led us into cosmological concerns. tem, but the whole point in considering a re- 23
24 theory and eigenforms. First, let us note « 3 » Heinz von Foerster, in discussing flexive system is that the agent, the observer, 24
25 that formally, mathematically, an eigenform what he called “eigenvalues” (Foerster 1981) is the system, and observers become both 25
26 is nothing more and nothing less than the and what I call “eigenforms” went off track the system and the parts of the system. Let 26
27 fixed point of a transformation in some do- in a carefully planned formal way that indi- the allegory become prose. The universe is 27
28 main. If the domain has name D and the cates a systematic abduction from the given the source of its own observation. The uni- 28
29 transformation is regarded as a function T: system into a larger context. He suggested verse is a self-excited circuit. The agents are 29
30 D → D, then an eigenform E is an entity (ei- considering the context-free application of not separate from their worlds. In §2, Fields 30
31 ther in D or in an extension of D) such that T upon itself, for any T whatsoever! And he et al. say that we propose an “epistemic cut” 31
32 T(E) = E. finds that he can take E = T(T(T(T(T(…))))) between agent and world for the purpose of 32
33 « 2 » Why do we take this notion of ei- and then with this infinite concatenation of theory construction. Theory demands such 33
34 genform to be of importance for cybernet- T upon itself, like the deep repeated reflec- a cut in order to distinguish a theorizing 34
35 ics? An initial answer is that the transfor- tions seen by an observer between two mir- agent. In fact, such a cut has to come along 35
36 mation T acting on a system D produces rors, we have T(E) = E. What has happened with any perception at all. And the key to 36
214 37 a natural recursion. Start with X(0), some here? Does this concept go too far? Any T the situation of perception is that we are 37
38 entity that we think may approximate a has a fixed point and that fixed point is noth- sensitive to the fact that while a distinction 38
39 fixed point. Let X(1) = T(X(0)). In general, ing more than an infinite reflection zone of is made, it is also mutable. There is no final 39
40 let X(n + 1) = T(X(n)) for n = 1, 2, 3, … ad copies of T in a circuit upon themselves. cut and in the acts of perception, as we come 40
41 infinitum. Then the transformation T be- Such a fixed point has no basis other than to our senses, we find those places of ambi- 41
42 comes the generator of a process and hence the transformation T itself. John Wheeler guity, of feeling, where it is not possible to 42
43 propels the system into time by the very ac- (Misner, Thorne & Wheeler 1973) had the say what is our construction and what is the 43
44 tion of the transformation. This process may same concept for quantum cosmology. He world. 44
45 have no fixed point. And we are well familiar said (in my paraphrase) that the Universe « 5 » In §3, the authors state: 45
46 with such a situation. In fact, almost every is a self-excited circuit, arising from its own 46
47 object or action that we know has a poten-
48 tially endless recursion associated with it.
observation of itself, which is that very ob-
servation of itself. There is nothing in the
“ We suggest that spacetime itself, including both
the space in which objects appear to be embedded
47
48
49 This applies in particular to fundamental universe except the self-participation of the and the time over which they appear to persist, is 49
50 transformations, such as simple motions of nothing that becomes information and form a relational, error-correcting code for the fitness 50
51 the human body like taking an upright step.
52 We take a step and we can take another. Of
arising from its own eternal return. The ei-
genform E is an existence and comes about
consequences of interactions.
” 51
52
53 course some transformations do have fixed in the cleft where spatial form and temporal At this point I am not prepared to com- 53
54 points. For example, T(x) = x2 has as a fixed process (time itself) meet. Von Foerster pro- ment on the nature of the code as error- 54
55 point the number 1, whose square is equal nounced this self-excited circuit in his own correcting. I am not clear what constitutes 55
column A column B column C

Constructivist Foundations vol. 12, N°3

 Second-Order Cybernetics
Questions Regarding Perception and Boundaries Konrad Werner

column A column B column C

1 an error in the context of the authors’ posi- gard them as eigenforms for the convenience them, the other’s possible observation is de- 1
2 tion. Thus, I ask the authors for clarification of the observer who is searching for deeper termined. This interrelationship goes across 2
3 about their notion of error (Q1). At one level understanding. Can the authors address this the structure of spacetime. This situation is 3
4 there can be no error. At another level, what issue from their point of view? (Q2) seen by some to be a paradox. I state it here 4
5 are called errors are certain distinctions « 7 » To offer support for my point in to bring into question the notion of Fields 5
6 made by an observer. At yet another level, §6, above, consider the fundamental situa- et al. that the fundamental source of the 6
7 errors are what are “corrected” by feedback. tion of the quantum mechanical model. The epistemic boundary is spacetime itself. The 7
8 In the case of eigenform, there can be “er- state of a quantum system is a vector in a Universe (now captitalized as we reenter the 8
9 ror correction” in the sense of stabilization complex vector space (a Hilbert space) that allegory) goes beyond spacetime and comes 9
10 if the recursive process does stabilize. But is seen as a sum over all possible observa- forth as self-excited circuit, living quantum 10
11 we also create forms of stabilization such as tions that can be made for the given experi- information, unified with a living observer 11
12 the infinite concatenation of the agent’s ac- ment. These possible observations are taken that is both distinct and not distinct from 12
13 tion in the form of E = T(T(T(…))). It is to as an orthonormal basis for the vector space, what is observed. 13
14 be understood that this infinite activity is an and the sum of the absolute values squared « 8 » I believe that it is a fundamental in- 14
15 abduction, a leap to a form that is invariant of the coefficients of the basis vectors is sight that Universe is identical with Self and 15
16 under T. It does not mean that an infinite equal to unity (now lower case). In this way, that it shall be possible to reformulate pres- 16
17 number of operations have been performed. the state vector is a probability distribu- ent-day physics so that it is seen as a form of 17
18 It means that the self-excited living circuit tion and indeed the probability of making the living. We are not there yet. Fields et al. 18
19 has come into being. In this sense, the eigen- an observation of one of these possibilities have gone forward with courage to explore 19
20 form corrects itself. It makes itself correct. I is equal to the absolute square of its coeffi- aspects of this possibility for Unity. 20
21 am the one who says I. cient. Physical processes are unitary trans- 21
22 « 6 » As in §12, the authors are quite formations that preserve the total probabil- Louis H. Kauffman is Professor of Mathematics at 22
23 taken with the notion of physical surface as ity distribution. Because we allow complex the University of Illinois at Chicago. He has a BS 23
24 the manifestation of the epistemic bound- coefficients, the superposition can model from MIT and a PhD in Mathematics from Princeton 24
25 ary. I agree that this is useful for physics interference and quantum effects. An ob- University. He has been President of the American 25
26 and particularly in the wake of the recent servation makes the distinction that brings Society for Cybernetics and is a Fellow of the American 26
27 holographic hypotheses in cosmology and forth one of the possibilities. This distinc- Mathematical Society and a recipient of the Warren 27
28 quantum physics. But the most generally ap- tion is often articulated without the usual McCulloch and Norbert Wiener medal of the American 28
29 plicable epistemic boundary is any distinc- spatial boundaries. Thus, a superposition Society for Cybernetics. His research is in cybernetics, 29
30 tion whatsoever. And when I say distinction, that indicates an entangled state does not form, topology, recursions and the theory of knots. 30
31 I mean an arising of observed difference and have to show the spatial structure that may 31
32 the arising of an observer of this apparent possibly separate the entangled particles. We Received: 7 June 2017 32
33 difference. I do not regard physical surfaces need only know the form of their entangle- Accepted: 14 June 2017 33
34 as fundamental epistemic distinctions. I re- ment to know that upon observing one of 34
35 35
36 36
37 37 215
38 Certain Questions Regarding get article by Chris Fields et al. very intrigu- from encoding the incoming data for us 38
ing. The text is insightful in many respects, and that this happens on the boundary that
39
40
Perception and Boundaries yet it is also dense, which makes addressing separates us from our surroundings. How-
39
40
41 Konrad Werner all the issues that should be addressed virtu- ever, crucially, the perceived spectacle does 41
42 ally impossible. For that reason, I shall focus not encode “information about the onto- 42
University of Warsaw, Poland
43 solely on the problem of boundaries. They logical or causal structure of the world, but 43
44
konrad.t.werner/at/gmail.com are salient factors in the proposal under rather information about the structure of 44
45 consideration, however, certain things need the fitness function that relates the agent to 45
46 > Upshot • I elaborate on how boundar- clarification. There is a literature on bound- the world” (§3). This means that the outside 46
47 ies are accounted for in the target ar- aries, in particular, a sub-discipline of ontol- world is a black box (a metaphor brought 47
48 ticle. This is a substantial issue if we are ogy called mereotopology (see references up several times by the authors themselves): 48
49 to understand the proposal laid out by in my target article in this issue). I shall not that it is, in a sense, hidden. 49
50 Fields et al. I argue that certain bound- refer to this literature, though, as it would 50
51 ary-related notions and theses need require much more space. How many boundaries? 51
52 clarification. « 2 » So, what does perception have to « 3 » Imagine that you are on a beach 52
53 do with boundaries? The authors say that and you see the line dividing the surface of 53
54 « 1 » Perception has to do with boundar- the perceived world, the familiar realm of the sea and the sky above it. Now, how many 54
55 ies. I find this general idea laid out in the tar- things that we perceive every day, results boundaries are there and to which entity 55
column A column B column C

http://constructivist.info/12/3/200.fields
 column A column B column C
1 do they belong? Is it the case that the water « 6 » So, suppose that we have one actly the information is encoded: on which 1
2 (or the air) is closed, meaning that it has its boundary shared by both sides as (1Bpw) boundary does this process occur? If it oc- 2
3 boundary as a part, while the air (respec- proposes. This means that both the perceiver curs on the perceiver’s boundary, then what 3
4 tively, the water) is open, i.e., it does not have and the world are closed, yet they are not role is left to be played by the world’s bound- 4
5 a boundary of its own, thus the boundary of separated. Imagine two pieces of a material ary? Perhaps here is the point where the idea 5
6 one entity serves as the boundary between sewn together: they are distinct and each of of the structure of fitness, as opposed to the 6
7 the entities in question? Or is it the case that them is bounded but they cannot be set apart; ontological structure of the world, comes 7
8 both the water and the air have their own they are parts of one whole, so to speak, pre- on stage. Suppose that the world’s boundary 8
9 boundaries and that these boundaries abut cisely because they are sewn. However, there provides a barrier that the perceiver bumps 9
10 each other? Finally, perhaps the water and is one subtle puzzle here: if the perceiver and against, so to speak, adjusting its shape, i.e., 10
11 the air share the boundary, meaning that the the world are sewn by their shared boundary, adjusting its boundaries, so that they fit, 11
12 latter is a common part of both entities. (I then one can hardly say that what happens in metaphorically, to the world’s boundaries. 12
13 omit the antirealist scenario in which there is the sewing itself has nothing to do with the However, if there are two separate boundar- 13
14 no boundary at all but only an illusion of its ontological structure of the world; after all, ies and their abutting determines the struc- 14
15 actual existence). These questions may seem this sewing is likely part of the ontological ture of fitness, then why is there any need for 15
Physics Concepts in Second-Order Cybernetics

16 silly, yet if being bounded is an essential fea- structure; if not, then what is it? a rather complex process of encoding infor- 16
17 ture of some entity, a condition of its identity, « 7 » In this context, we can notice an mation and establishing this whole theater 17
18 the issue becomes ontologically critical. interesting tension in the very nature of at of phenomena that we face once we open 18
19 « 4 » When it comes to our case, the least some boundaries. Think of a living our eyes in the morning? This is just another 19
20 authors claim that perception is a spectacle creature: boundaries constitute an organ- way of formulating what David Chalmers 20
21 played on the boundary between the per- ism by cutting it off from its environment, (1995) once called the hard problem of con- 21
22 ceiver and the outside world (the black box), yet at the same time, they provide channels sciousness, yet from a different side; this is, 22
23 but how many boundaries do we have there? for communication with the environment. say, the hard problem of presentations: why 23
24 Here is the first option: Say, once they bound something, they open there presents something rather than noth- 24
25 some doors to make traffic possible. When ing; why are we not “zombies,” bumping 25
26 (1Bpw) There is one boundary. It belongs it comes to the philosophy of mind and against the boundary of the world, adjusting 26
27 to the perceiver and to the outside world; it perception this tension is crucial: there is to it and by doing so maintaining solely our 27
28 is shared by them. They are both closed. the Cartesian approach to the mind-world structure of fitness? It seems that we could 28
29 boundary, putting stress on isolation or do so without facing any phenomena and it 29
30 The perceiver The outside world separation, while, e.g., in Edmund Husserl’s is likely that the most primitive organisms 30
31 or Maurice Merleau-Ponty’s approaches, the still function in this way. 31
32 boundary in question was supposed to – let « 9 » Perhaps stripping the perceiver 32
This option is clearly endorsed by the au-
33 me use Husserl’s original and very pregnant from its boundaries yields an even better 33
34 thors in §10. formulation – bring the world to a presenta- understanding of the structure of fitness. 34
35 « 5 » If we generalize what the authors tion. The authors apparently take the Carte- 35
36 say in §10, then the perceiver and the out- sian route and I am not sure if that is neces- (1Bw) There is one boundary and it be- 36
216 37 side world are neither closed nor open (origi- sary for their project as a whole. longs to the outside world. The perceiver is 37
38 nally, the model outlined in the target article « 8 » But perhaps there are actually two open while the world is closed. 38
39 referred to the simplified situation in which boundaries, as it is also suggested in §10, 39
40 an outside world for one perceiver is another where the authors introduce a distinction The perceiver The outside world 40
41 perceiver). This cannot be right. The alleged between a boundary and its surfaces. But is 41
42 “purely notional” character of the bound- a surface not a boundary, too? So, we can at 42
43 ary in question has nothing to do with the least take into consideration the following Here the perceiver is shaped by the bound- 43
44 context. This is because, by deeming the scenario: aries of the world as boundless water poured 44
45 boundary “purely notional” we take a par- into the glass. In this sense, the perceiver fits 45
46 ticular position as regards the nature of the (2B) There is one boundary that belongs the boundary (or boundaries) of the world. 46
47 boundary, not about its very existence. So, for to the perceiver and one boundary that This boundary must be there, pre-given and 47
48 example, the boundary between Poland and belongs to the outside world. They are ready-made (Hilary Putnam’s term) inde- 48
49 Russia (Kaliningrad Oblast) is purely con- both closed. pendently of the perceiver if the latter is sup- 49
50 ventional, but surely it exists and it is even posed to adjust itself to it. Such a scenario 50
51 guarded by heavily armed forces. So, if there The perceiver The outside world has been discussed and criticized, e.g., by 51
52 is a boundary (or boundaries) between the Francisco Varela, Evan Thompson and Elea- 52
53 perceiver and the outside world, regardless nor Rosch (1991: 193, 198). However, aside 53
54 of its nature, the two realms must be either However, this case is very problematic due from Varela’s criticism, here as in the (1Bpw) 54
55 open or closed. to the fact that it becomes unclear where ex- scenario, it is not clear why the structure 55
column A column B column C

Constructivist Foundations vol. 12, N°3

 Second-Order Cybernetics
Questions Regarding Perception and Boundaries Konrad Werner

column A column B column C

1 of fitness is distinct from the alleged onto- suggest that there is just one place where came into being, so that, say, the vision- 1
2 logical structure of the world. After all, the the perceiver-world boundary is drawn so determined boundary of the perceiver is 2
3 boundary that the perceiver faces partakes that both sides seem to be like two blocks. not identical to its physical boundary qua 3
4 in this ontological structure. If it does not But what forces us to accept this two-blocks organism, and at the same time this new 4
5 partake in it, then what is it? Thinking of the model? Maybe it would be much better to boundary serves as the vision-determined 5
6 boundary in question as if it were like a mere draw several lines composing a more com- boundary of the thing perceived. Here per- 6
7 wrapping paper having nothing to do with plex structure, say, something like this: ception, cognition in general, brings forth 7
8 the thing being wrapped – the world in this significantly new types of boundaries and 8
9 case – makes the boundary a mysterious, su- – this is a constructivist aspect of the idea 9
10 perficial entity of unknown origin. But if it is The perceiver – imposes these boundaries on the world 10
11 not a mere wrapping paper, then one cannot so that the world is brought to a presenta- 11
12 say that the ontological structure is hidden The outside world tion in such and such a guise (see my target 12
13 behind the boundary; the structure is there, article in this issue). And perhaps further 13
14 and the boundary is its manifestation. steps in this evolutionary process resulted 14
15 « 10 » Finally, there is a scenario that in the boundaries of what we used to call 15
16 strips the world from its boundary: « 13 » Someone might say that there is mind. Recall Andy Clark and Chalmers’s 16
17 no essential difference between the latter (1998) groundbreaking idea of an extended 17
18 (1Bp) There is one boundary and it be- and former scenarios. While this is correct, mind. What they propose boils down to the 18
19 longs to the perceiver. The world is open the latter drawing makes an important sug- claim that the mind sets its special arrange- 19
20 while the perceiver is closed. gestion: both sides, i.e., the perceiver and ment of boundaries that are not identical to 20
21 the world, are shaped with respect to each the physical boundaries of the body. 21
22 The perceiver The outside world other; the boundary line is not just a line; 22
23 it contributes to what the two bounded Konrad Werner works at the University of Warsaw, 23
24 realms are. Poland, where he is running a research project in 24
25 Varela et al. likely have this scenario in « 14 » There may be, however, an es- experimental philosophy. He obtained his PhD in 25
26 their minds when they write that “our lived sential difference, too. The essential, yet 2013 from the Jagiellonian University in Krakow. He 26
27 world does not have predefined boundar- rather tacit assumption behind what I have is interested, first of all, in so-called “aspectual 27
28 ies” (Varela, Thompson & Rosch 1991: just dubbed provisionally the two-blocks shape” or perspective-dependency of perception and 28
29 148), and model is that there is just one type of per- knowledge. His articles concern epistemological, 29
30 ceiver-world boundary. But why? Perhaps ontological as well as logical aspects of this issue. 30
31
“ cognition is not the representation of a pre-
32 given world by a pregiven mind but is rather the
each perceptual subsystem, be it vision,
hearing or touch, sets up and imposes on
Still in connection with it, he has recently directed his
attention toward evolutionary origins of cognition.
31
32
33 enactment of the world and a mind on the basis of the world its own structure of boundaries. 33
34 a history of the variety of actions that a being in Note that from the evolutionary perspec- Received: 19 May 2017 34
35 the world performs.
36
” (ibid: 9). tive, the step from mere mechanical senses
like touch or from chemical senses, the
Accepted: 20 June 2017 35
36
37 Here, admittedly, we cannot speak of a pre- oldest ones, to vision – the step that marks 37 217
38 given or ready-made ontological structure a great evolutionary achievement – origi- 38
39 of the world in the absence of what the per- nated from a new ability to target what was 39
40 ceiver does. However, whatever the struc- literally on the boundary of an organism, 40
41 ture of this world is, it cannot be regarded as where receptors are plugged in, not as the 41
42 obscured or hidden either. Here, structures object perceived but as a signal of an object 42
43 of the world result from the perceiver’s in- or as information. Hence, while in the case 43
44 teractions with the world. of touch, the boundary of the thing being 44
45 « 11 » To conclude this part, I wonder perceived abuts the physical boundary of 45
46 to what extent the conception outlined in the perceiver (let alone chemical sensation 46
47 the target article could be freed from the where a substance that is perceived must 47
48 Cartesian idea of the world’s being ready- react with certain proteins, which makes 48
49 made (its having a structure independently the question of boundaries difficult – there 49
50 of cognition) and hidden (meaning the in- is something more than abutting), in the 50
51 accessibility of this ready-made structure). case of vision, for instance, these respective 51
52 boundaries have nothing to do with each 52
53 How many types of boundaries? other. But perhaps – let me set this off as 53
54 « 12 » The scenarios presented above, as a speculative hypothesis – together with 54
55 well as those presented in the target article, vision, a specific new system of boundaries 55
column A column B column C

http://constructivist.info/12/3/200.fields
 column A column B column C
1 “Eigenforms, Interfaces and decoherence (§§15–18). The underlying as- cently underwent a popular revival in the lit- 1
sumption is therefore that a theory of per- erature of high energy physics, thanks to the
2
3
Holographic Encoding”: Their ception should be addressed and studied intuitions of Stephen Hawking (2015) and to
2
3
4 Relation to the Information through the lenses of quantum mechanics. the work of Hawking, Malcolm Perry and 4
The authors explicitly mention criticism of Andrew Strominger (2016) on soft photons.
5
6
Loss Paradox for Black the assumptions of “epistemic transparency The very quantum-mechanical description
5
6
7 Holes and Quantum Gravity and objective persistence” proper of the of the theory of perception the authors move 7
8 classical worldview and point toward the from can naturally encode quantum hairs. 8
9
Antonino Marcianò elaboration of experiences within the theo- But then a first provocative question to ad- 9
10 Fudan University, China retical framework of quantum mechanics. dress would be: Can an observer have access 10
11 marciano/at/fudan.edu.cn They ascribe particular relevance to unitary to quantum hairs and thus to the informa- 11
12 quantum theory (§15) as the correct para- tion that can be encoded in these latter enti- 12
13 > Upshot • I emphasize possible analo- digm in which to address decoherence and ties? (Q1) 13
14 gies and links between the content of holographic encoding. As they point out, « 4 » Beyond this analogy between 14
15 Fields et al.’s target article and some in unitary quantum theory “the universe is event horizon and interface of perception, it 15
Physics Concepts in Second-Order Cybernetics

16 consolidated recent studies in the lit- permanently in an entangled state; there are is possible to point out a more general cor- 16
17 erature of quantum gravity and the in- no classical objects” (§15; emphasis in the respondence between quantum degrees of 17
18 formation loss paradox for black holes. original). freedom that are encoded on the observer- 18
19 This follows from the attempt by the « 3 » Besides the philosophical prefer- system boundary and some theories of quan- 19
20 authors to account for spacetime as an ence toward unitary quantum mechanics, tum gravity that make explicit use of graph 20
21 error-correcting code. The paradigm the the line of thought followed by the authors theory. Among the latter we mention loop 21
22 authors focus on can be naturally cast in has a striking overlap with a vast part of the quantum gravity (Rovelli 2004) and theoreti- 22
23 the language of some models of quan- literature developed in the last four decades cal constructions that arise from string-nets 23
24 tum gravity based on graph theory, and about the information loss paradox of black (Levin & Wen 2005). Gauge interactions, 24
25 suggests a generalization of the percep- holes, and crosses its natural consequence, and eventually also fundamental particles 25
26 tual systems so as to account for quan- which is the development of the holographic of the standard model (Bilson-Thompson, 26
27 tum holographic encoding as described principle – see, e.g., the seminal works by Markopoulou & Smolin 2007), can be de- 27
28 in quantum gravity. Gerard ’t Hooft (1993) and Leonard Suss- rived in these two frameworks. The basic 28
29 kind (1995) – in quantum gravity and high objects of these theories are graphs, namely 29
30 « 1 » At the core of the target article energy physics (Bousso 2002). There are evi- sets of nodes interconnected by links, which 30
31 “Eigenforms, interfaces and holographic dent analogies that assimilate the crucial role are colored by fundamental representations 31
32 encoding: Toward an evolutionary account of the black hole event horizon in the flow of some continuous or discrete Lie group. 32
33 of objects and spacetime” there is the de- of information to the role of the membrane These latter are sets of elements on which it 33
34 velopment of the interface theory of per- between observer and system in the inter- is possible to define a product rule, recover 34
35 ception (Hoffman, Singh & Prakash 2015). face theory of perception. The comprehen- a unit element and then find an inverse el- 35
36 This framework is unfolded within the sion of the function of the physical degrees ement that reproduces the unit element by 36
218 37 very same language in which the epistemic of freedom that puncture the observer–sys- virtue of the product rule. The redundancy 37
38 foundations of quantum mechanics can be tem interface represents a possible pathway that the authors propose to be deployed for 38
39 phrased (§§4–13). The interface theory of to solve the information loss paradox. The unravelling the emergence of space, and in 39
40 perception allows a detailed description of key point is to overcome the no-hair theo- general spacetime, as an error-correcting 40
41 the holographic encoding, and is naturally rem for classical black holes. This states that code could be then associated with the ir- 41
42 tailored in order to account for the com- the thermodynamics of black holes shall be reducible representations that are assigned 42
43 plexity of the observer–environment inter- described only in terms of three quantities: to the links of the graphs in these theories. 43
44 face’s interactions. Within this framework the mass, the spin and the electric charge of This is exactly the same construction de- 44
45 the authors address the structure itself of black holes, all the other classical degrees of veloped in loop quantum gravity or string- 45
46 spacetime (§§32–36), after having reviewed freedom being irrelevant (Misner, Thorne & nets. The quantum states of the models of 46
47 and analyzed the most relevant options for Wheeler 1973). The no-hair theorem can be emergent spacetime are then recovered from 47
48 holographic encoding (§§15–18), and sum- avoided by resorting to the notion of quan- the graphs that are taken into account. The 48
49 marized the propositions for the fitness tum hairs. The latter are quantum numbers colors, i.e., the irreducible representations of 49
50 functions (§§22–24) deployed in the inter- that black holes may carry, which are not elements of the Lie group, are now associated 50
51 face theory of perception. associated with massless gauge fields and with eigenvalues of the observable quantum 51
52 « 2 » Although the axioms of quantum which may solve the information paradox, operators of the theory. The dimension of the 52
53 mechanics are not explicitly stated, the focus allowing for storing of information. This is Hilbert space associated with the irreducible 53
54 throughout the work is on quantum states, a perspective that comes from an old idea representations of a discrete or continuous 54
55 entanglement and observer-environment (Coleman, Preskill & Wilczek 1992) that re- group Lie group G – or eventually to a quan- 55
column A column B column C

Constructivist Foundations vol. 12, N°3

 Second-Order Cybernetics
Boundaries, Encodings and Paradox Antonino Marcianò

column A column B column C

1 tum group – that are assigned to the links of and consistently describe at the quantum(- Authors’ Response 1
the graphs, is the natural instantiation of this gravitational) level the interactions these un-
2
3 line of thought and might represent redun- dergo. The role of the symmetries, to which
Boundaries, Encodings and 2
3
4 dancy. are connected charges that may play the role Paradox: What Models Can 4
« 5 » Moving from such an intuitive ap- of bits, is indeed very intertwined with this
5
6 proach we are led to ask a second relevant aspect, as emphasized in a series of studies by
Tell Us About Experience 5
6
7 question: How can we encode the observer- Hawking, Perry and Strominger — see, e.g., Chris Fields, Donald D. 7
8 system duality in those models of quantum Hawking, Perry & Strominger (2016).
Hoffman, Chetan Prakash 8
9 gravity that are phrased within the language « 8 » I wish also to emphasize that the 9
10 of graphs (spin-networks, Wilson loops or role of quantum gravity is not only crucial to & Robert Prentner 10
11 string-nets)? (Q2) If we take into account determine the dimensionality of the bound- > Upshot • Formal models lead beyond 11
12 the authors’ analysis in (§§15–17), it seems ary Hilbert space – this pertains to the total ordinary experience to abstractions such 12
13 natural to argue that the individuation of an set constituted by the “observer” and the as black holes and quantum entangle- 13
14 interface distinguishes two subsystems of the “system,” and accounts for the description of ment. Applying such models to experi- 14
15 Hilbert space, and thus implies that the total their interaction – but is relevant as well to ence itself makes it seem unfamiliar and 15
16 set of degrees of freedom encloses both the regularize the maximum amount of degrees even paradoxical. We suggest, however, 16
17 observer that perceives and the perceived of freedom that shall be considered while that doing so also leads to insights. It 17
18 system. Nonetheless, redundancy would re- reckoning the exchange of bits and the flow shows, in particular, that the “view from 18
19 quire, in order to let emerge the notion of of information through the membrane. This nowhere” employed by the theorist is 19
20 spacetime that satisfies the Einstein equiva- provides a set-up in which we can operation- both essential and deeply paradoxical, 20
21 lence principle, that a continuous group ally accomplish calculations, avoiding infini- and it suggests that experience has an 21
22 structure G could be consistently defined ties. unrecorded, non-reportable component 22
23 (§§15–17) for the spaces of actions, and that « 9 » I end this brief commentary by in addition to its remembered, report- 23
24 this G could be connected to the Lorentz recalling the authors’ suggestive remark (in able component. 24
25 group. §38) – part of common belief in the commu- 25
26 « 6 » As the authors point out in §34, a nity of quantum gravity that has been grow- « 1 » We thank our commentators for 26
27 sizable amount of energy expenditure is re- ing in recent years – that with the relation be- their insightful criticism. While each of them 27
28 quired for the holographic encoding, which tween interface’s perception and holographic chooses a different focus for their comments, 28
29 is roughly proportional to the number of bits encoding we may only actually be probing the issues they raise overlap considerably. We 29
30 involved at the interface and to thermal en- the tip of an iceberg. A deeper understand- highlight in what follows what we take to be 30
31 ergy for each degree of freedom. This implies ing of the emergent nature of spacetime the major issues, and attempt to show how 31
32 that the redundancy increases in propor- might indeed arise from the development they relate both to what we propose in the 32
33 tion to the dimension of the Hilbert space at of a theory of quantum information gravity target article and to one another. 33
34 the boundary between the observer and the that many authors are currently developing 34
35 system. Thus, the simple system described in the literature. The “classical world” is the 35
36 by a binary code, namely the Hilbert space explanandum 36
37 of spin 1/2 particle in the physicist’s jargon, Antonino Marcianò joined as Associate Professor « 2 » Constructivists, phenomenolo- 37 219
38 might already turn into an extremely com- the Department of Physics at Fudan University in gists, and others who reject naive realism 38
39 plicated model to be solved. Nonetheless, January 2014, becoming a member of the theory are faced with the task of explaining a shar- 39
40 at least from a theoretical perspective, we and high-energy division. Previously a post-doctoral able experience of a classical world – a world 40
41 may ask what happens if the Hilbert space researcher at Princeton University and Dartmouth of “tables and chairs and dogs and cats and 41
42 at the boundary is composed by N degrees College, he was studying models for cosmological people” (Eric Dietrich §1). Even the “natu- 42
43 of freedom whose internal degeneracy is inflation and CMBR physics, currently his main topics ralized” sciences, however, face this chal- 43
44 described by the irreducible representation of research. In the USA, he also continued focusing on lenge. This is obvious in the case of quan- 44
45 of a Lie group G. The main last question I the Wilson-loop approach to Quantum Cosmology and tum theory, but even the classical theory of 45
46 propose is therefore: What is the nature of Quantum Gravity, learnt while working at Aix-Marseille atom-based matter – the classical physics of 46
47 these degrees of freedom at the interface be- University, soon after his PhD at Sapienza University the late 19th century – faces the problem of 47
48 tween the observer and the system and what of Rome. His current research also encompasses how clouds of atoms could appear to us to be 48
49 is the internal degeneracy group, namely the the implementation in condensed-matter physics of tables or chairs. It is less obvious in the case 49
50 redundancy, connected to these degrees of mathematical tools borrowed from quantum gravity, as of biology and psychology, but here it must 50
51 freedom? (Q3) an attempt to address dynamics on lattice structures, be explained how agglomerations of cells – 51
52 « 7 » The answer to Q3 amounts to the including graphene, in non-perturbative regimes. i.e., organisms – could self-assemble in ways 52
53 correct reconstruction of the boundary that allow the experience of such things as 53
54 physical theory. We must indeed recover the Received: 19 June 2017 tables and chairs as opposed to, say, just 54
55 relevant degrees of freedom at the interface, Accepted: 26 June 2017 brightness and saltiness. 55
column A column B column C

http://constructivist.info/12/3/200.fields
 column A column B column C
1 « 3 » We agree with Dietrich (§14) that Ernst von Glasersfeld puts it, summarizing clear how anyone could speak one particular 1
2 the experience of a classical world is in- three millennia of philosophical empiri- sentence or think one particular thought. 2
3 eluctable. When we open our eyes, we see cism, “it is impossible to compare our im- Memory and communication both become 3
4 bounded objects with definite shapes, sizes age of reality with a reality outside” (Gla- paradoxical. Any non-classical theory seems 4
5 and locations; when we open our ears we sersfeld 1981: 89). When we imaginatively to require, as Niels Bohr argued, a classical 5
6 hear tones with definite loudness and pitch. construct theories of what lies beyond the metatheory just to support language. Here 6
7 Our goal is to explain why we have such interface, we construct and express them a dialetheic world (Dietrich §12) seems ines- 7
8 experiences. Dietrich suggests that the ex- using symbols and diagrams that our inter- capable (Dietrich & Fields 2015). 8
9 perience of a classical world is ineluctable faces allow: classical symbols and diagrams « 7 » While we do not, as Dietrich points 9
10 because there is an ontologically real clas- that have definite arrangements and shapes. out, have direct empirical evidence for our 10
11 sical world, one with a “mind-independent Such symbols and diagrams are, like our model, there is plentiful (albeit indirect) evi- 11
12 spacetime” that is “filled with mind-inde- percepts, eigenforms, fixed points that are dence for holography as a mechanism (see, 12
13 pendent objects” (§2). We “visit” this world only recognizable through repeated use. We e.g., Antonio Marcianò’s commentary). Many 13
14 by opening our eyes and ears. According to have no choice in our use of classical sym- would argue, moreover, that the mounting 14
15 Dietrich (§12), an utterly differently struc- bols and diagrams, as our experiences of evidence for quantum effects at macroscop- 15
Physics Concepts in Second-Order Cybernetics

16 tured quantum world that we can access theory construction and our experiences of ic scales demonstrates empirically that the 16
17 (since the 1920s) only via our thoughts can our constructed theories are experiences and classical worldview is wrong. As Dietrich em- 17
18 be considered to be equally real, and there so are encoded on our interfaces. The classi- phasizes, accepting this argument requires 18
19 may be other equally real worlds with yet cal symbols and diagrams that we use to ex- the acceptance of another deep paradox. 19
20 different structures that we cannot access at press our theories make use of redundancy Experiments, in particular, require time- 20
21 this time. From a constructivist perspective, in space and time; hence they enable error persistent observers and apparatuses that 21
22 these “worlds” are all constructs, one of our correction. interact while remaining separable in the 22
23 perceptual systems and the other(s) of our « 6 » What we have called the classical physicist’s sense of having independently 23
24 theoretical imaginations. Why the former worldview, on the other hand, is an assump- characterizable states. Joint states of inter- 24
25 should provide compelling experimental ev- tion that the classical world of our experi- acting systems are not, however, separable 25
26 idence for the latter remains a mystery. Why ence is not just encoded on our interfaces, under the unitary evolution prescribed by 26
27 we can only express our theories – even to but also exists beyond them as an ontologi- quantum theory. This paradox can be stated 27
28 ourselves, in thoughts – using classical sym- cally real structure comprising a multitude starkly: local decoherence requires global 28
29 bols is also mysterious. of well-defined, bounded, time-persistent coherence, i.e., global entanglement. From a 29
30 « 4 » We attempt to address these ques- macroscopic objects. We see tables and global quantum-theoretic perspective, both 30
31 tions by appealing to a specific mechanism: chairs, in this worldview, because tables decoherence and the classical world it pro- 31
32 holographic encoding on an interface that and chairs (not just clouds of atoms) are out duces are epiphenomenal. 32
33 employs spacetime as an error-correcting there, bouncing light into our eyes. Percep- « 8 » Dietrich also points out (§14) that 33
34 code. We (each) see a classical world, in our tion is (mostly) veridical because the in- we have offered no theory of how human 34
35 view, because we (each) have this kind of terfaces through which we have perceptual beings can formulate, within their classical 35
36 interface. The “objects” – including objects experiences are (mostly) transparent. The interfaces, theories of the non-classical. This 36
220 37 of thought – that our interfaces present to world, on this worldview, is not a black box is a fair challenge that we hope someday to 37
38 us are eigenforms. As Heinz von Foerster at all, but rather a (mostly) white one. What accept. 38
39 (1976) emphasized, eigenforms and the cor- you see is what you get. Dietrich argues (§3, 39
40 responding eigenbehaviors are (at least ap- §12) that this world/worldview distinction Consciousness is fundamental, but 40
41 proximate) fixed points of multiply repeated is illegitimate without empirical evidence architecture must be fundamental 41
42 (ideally infinitely repeated) perception-ac- that our model is correct. We disagree: the too 42
43 tion loops (cf. Louis Kauffman’s commentary). classical worldview is an explicit philosophi- « 9 » Both Dietrich (§13) and Urban 43
44 Eigenform and eigenbehavior must be clas- cal claim or, more commonly, an implicit Kordeš (§10) suggest that we are trying to 44
45 sically correlated across these repetitions; and perhaps innate assumption that can be explain phenomenal consciousness, or are 45
46 hence the process of repetition, whether it (and in point of fact is) made independently at any rate not taking it to be fundamental. 46
47 is conscious or not, constitutes a memory. It of whether the classical world that it postu- We were perhaps not sufficiently clear that 47
48 is this memory of classical correlation that lates actually has the ontological status that we take phenomenal consciousness to be 48
49 confers classicality on the “classical world” the classical worldview claims it to have. On fundamental and irreducible, and simply as- 49
50 of our interface-encoded experience. the other hand, we agree with Dietrich that sume that conscious agents have it. However, 50
51 « 5 » If we are correct, the “classical there is a deep issue here: stating this dis- we also assume that conscious agents have 51
52 world” is not a world at all, but is only an tinction is making a statement, and making an architecture in addition to consciousness. 52
53 experience. The classical-world experience any particular statement is a classical act. If The structure and content of phenomenal 53
54 is ineluctable because the interface that the classical worldview is rejected, the sta- consciousness (i.e., experience) alone is, we 54
55 encodes it is the only interface we have; as tus of statements is cast into doubt; it is un- claim, insufficient to explain itself, e.g., in- 55
column A column B column C

Constructivist Foundations vol. 12, N°3

 Second-Order Cybernetics
TOC Antonino Marcianò

column A column B column C

1 sufficient to explain the structure and con- conscious realism (see §9 in our target ar- imes with more or fewer dimensions or even 1
2 tent of the experienced classical world. ticle). This postulate is not as radical as it different geometries from ours. Encodings 2
3 « 10 » Kordeš specifically argues that our seems. Two CAs defined to have the same of some kinds of human experience, e.g., of 3
4 distinction between the experience space “world” set W can be taken to represent two emotions or epistemic feelings, tend to em- 4
5 X and the space G of available actions is a “points of view” on W. If, however, W is ploy time but not space. Nothing requires or 5
6 mistake; G, Kordeš suggests, should be a sub- reconceptualized as simply the information even suggests a common encoding across 6
7 set, presumably a proper one, of X. “Being channel via which the agents interact, its de- the entire interface. 7
8 autonomous,” he claims, “means that the grees of freedom can be subsumed into the « 14 » The notions of open and closed 8
9 agent chooses from the options the agent perception and action maps of the agents boundaries of classical mereotopology are 9
10 itself constructs rather than from pre-given to produce the interacting-agent configura- motivated by the characteristics of ordinary 10
11 options” (§11). Placing G within X results, tion shown in Figure 1c of our target article. objects occupying continuous, locally Eu- 11
12 however, in an agent aware of every available From the perspective of either agent, the clidean spacetime. Hence it is unsurprising 12
13 action and of every choice of action. No ac- “world” is indistinguishable from the other that, as Werner shows, they are of little use 13
14 tions by such an agent can be “automatic” as agent. René Descartes realized this in his in understanding the kind of interface pro- 14
15 psychologists such as John Bargh and Tanya Meditations, stating that nothing in his ex- posed here. Werner rejects, in particular, our 15
16 Chartrand (1999) use this term. Genuine perience could prove that he was not inter- characterization of observer (or “perceiv- 16
17 autonomy, moreover, requires that the agent acting with an “evil demon” that synthesized er”) and environment (“outside world”) as 17
18 be able to actually perform whatever action his every percept. The currently fashionable mereotopologically neither open nor closed 18
19 is chosen. This is possible only if the world idea that we (each) live in a computer simu- (Werner §5). If either is open, its complement 19
20 never interferes to prevent a chosen action. lation constructed by some advanced race, must be closed (Smith 1996). Observer and 20
21 The conscious agent (CA) formalism sepa- maybe even our own descendants (Bostrom environment are, however, on this model 21
22 rates G from X not just to enable automa- 2003), updates Descartes. The simulation is, entirely equivalent and interchangeable; this 22
23 ticity, but also to take the evident ability of in this view, the channel by which the aliens, is why we draw them symmetrically and 23
24 the world to interfere with our desires into or maybe our grandn-children, toy with us. prefer the neutral “Alice” and “Bob” nomen- 24
25 account. The best argument for the existence clature to the connotation-laden “observer” 25
26 of a world independent of your own mind is, The interface is a boundary in state and “environment.” Nothing motivates any 26
27 as The Rolling Stones explain it, “you can’t space, not spacetime structural distinction between the two; 27
28 always get what you want.” « 13 » Kauffman and Konrad Werner both hence there is no justification for a mereo- 28
29 « 11 » Postulating an architecture is, by wonder how the interface is defined, a ques- topological distinction. Given that they 29
30 its very nature, going beyond “lived expe- tion that is present but implicit for both Di- interact, we are left with the situation that 30
31 rience” to the realm of theoretical models. etrich and Kordeš. Kauffman asks, in particu- Werner (§4) labels “1Bpw”: both systems are 31
32 We fully agree with Kordeš that pretending lar, (Q2) whether we require the interface to closed and they share a boundary. While the 32
33 to “eyes of God” that “[see] all agents, their be a “physical surface,” later attributing to us boundary is shared, however, the systems 33
34 actions and interactions” (§13) is a mis- the notion that “the fundamental source of cannot both be closed: observer and envi- 34
35 take, but we nonetheless regard an ability the epistemic boundary is spacetime itself ” ronment together compose the entire uni- 35
36 to build, consider, and derive predictions (§7). The word “physical” here is ambigu- verse, which, as Barry Smith (1996) points 36
37 from theoretical models as an essential ad- ous; physicists often use it to mean merely out, is boundaryless and hence not mereo- 37 221
38 junct to phenomenology. The formalism “consistently describable in the language of topologically closed (it is, however, closed in 38
39 and diagrams of von Foerster, for example, physics,” ruling out as “unphysical” only sit- the physicist’s sense of not interacting with 39
40 compose such a model, as do those of Karl uations with mathematical descriptions that anything). This situation is rendered even 40
41 Friston or Wojciech Zurek or indeed of any are self-contradictory or meaningless. We more paradoxical by noting that observer 41
42 other author who claims to explain or pre- can, however, state categorically that we do and environment each appears fully embed- 42
43 dict any experience of any observer. Kordeš is not require the interface to be a boundary in ded in the other when viewed from their 43
44 no exception. “By renouncing the view from spacetime, and we apologize if anything in own perspective. 44
45 nowhere, consciousness appears every- our text suggests this. We regard spacetime « 15 » Kauffman remarks that “the most 45
46 where” (Kordeš §15) may well be a report of as a way of encoding information on an in- generally applicable epistemic boundary is 46
47 first-person experience, but saying how this terface, one that may or may not be used, but any distinction whatsoever” (§6). The dis- 47
48 happens requires a model. For many, more- that provides the benefit of some level of er- tinctions between red and green or between 48
49 over, consciousness appears everywhere ror correction. Human experience and thus happy and sad are examples. Any prop- 49
50 only from a theoretical, view-from-nowhere the (typical) human interface employs spa- erty that supports such a distinction (what 50
51 perspective, one from which the futility of cetime to advantage for encoding percepts, physicists call a “degree of freedom”) can be 51
52 attempts to make consciousness “emerge” some concepts (e.g., those of geometry), and thought of as a component of the state of a 52
53 from something else becomes evident. much of what we imagine, but other kinds system. The boundaries in which we are in- 53
54 « 12 » Consciousness appears every- of observers may have interfaces that do not terested are boundaries in the abstract state 54
55 where in the CA framework via a postulate: employ spacetime, or that employ spacet- space (as Kauffman §7 points out, this is a 55
column A column B column C

http://constructivist.info/12/3/200.fields
 column A column B column C
1 Hilbert space in quantum theory) of the uni- & Jekníc-Dugíc 2008; Harshman & Ranade verse, consciousness and its contents are, 1
2 verse. Observer and environment are distin- 2011); this constancy of whole-system dy- like decoherence, epiphenomenal. Decom- 2
3 guished as subsystems by the states that they namics under arbitrary decomposition has positional equivalence renders a universe 3
4 can occupy. The epistemic boundary be- been termed “decompositional equivalence” filled with awareness and a universe con- 4
5 tween them – the boundary by which we, as (Fields 2016). Within the CA formalism, taining no awareness indistinguishable from 5
6 theorists, distinguish them – is their shared decompositional equivalence is imple- a (theoretical) perspective that stands “out- 6
7 interface. The states on this boundary are mented by the arbitrary composability of side” of it. The “view from nowhere,” even 7
8 available to encode experiences; they imple- Markov processes. The universe as a whole when adopted via an abstract model, is in- 8
9 ment the respective spaces X of observer and has no “outer” boundary; decompositional herently paradoxical. 9
10 environment in the CA formalism. What is equivalence allows the erasing of any “in- 10
11 encoded on the interface at any instant of ner” boundaries as well. Hence the universe Experience is both classical and 11
12 either system-relative time depends on how can be considered to be filled with observ- non-classical 12
13 the two systems are interacting at that time. ers and experiences as described above, but « 20 » A partial resolution of this para- 13
14 The interaction need not involve spatial de- the boundaries defining these observers can dox of disappearing awareness may come 14
15 grees of freedom, as Kauffman makes clear in also be erased with no effect. In the CA for- from an unlikely corner. Marcianò focuses on 15
Physics Concepts in Second-Order Cybernetics

16 his discussion of entanglement (§7). malism, the universe can be considered to a particular system for which the state-space 16
17 be a CA or any combination of CAs, but it boundary corresponds to a spatial bound- 17
18 All boundaries encode experience, can also be considered to be a single set W ary, the black hole, and asks (Q1) how our 18
19 but all boundaries can be erased mapped to itself. If any distinction creates a approach might deal with the paradox that 19
20 « 16 » Kauffman’s remark that “any dis- boundary, such a boundaryless system can black holes appear to destroy information 20
21 tinction whatsoever” creates an epistemic make no distinctions. With no boundary to whenever they gain energy, in violation of 21
22 boundary is, however, even more powerful serve as an interface and no ability to make quantum theory’s requirement of unitar- 22
23 than this. It implies, when taken seriously, distinctions, the universe has no experience ity and hence information conservation. As 23
24 that every possible boundary in state space space X and no experiences. It has no point Marcianò points out, one answer to this para- 24
25 encodes experience. Every system is an ob- of view, on itself or on anything else. John dox is to recognize that black holes are only 25
26 server; likewise, every system is an observed Wheeler’s well-known statement (Kauffman apparently classical objects; they are entan- 26
27 environment. Every state corresponds to an §3) is, therefore, misleading. The universe gled with the rest of the universe by “soft” 27
28 experience on some interface. The universe is composed of observer-participants, but is, photons and possibly other “quantum hair” 28
29 is, therefore, filled with experiencers and when viewed as a boundaryless whole, itself (see Strominger 2017 for a recent elabora- 29
30 filled with experience. In this sense, contra neither an observer nor a participant. tion of this view). 30
31 Kordeš (§17), the abstract space in which « 18 » Taking actions into account deep- « 21 » As all systems smaller than the 31
32 agents live is indeed a space of phenomenal ens the above paradox. Boundaries encode universe as a whole are observers in our 32
33 experience. Each agent, however, experienc- not just experiences but actions: the per- approach, black holes are observers. In- 33
34 es only what is encoded on its own interface. ceptions of each agent are the actions of its deed, they are ideal observers: all informa- 34
35 Sensations, thoughts, feelings, imaginations, environment and vice versa. The actions of tion (particles or waves) that contacts their 35
36 the experiences of deciding or doing, all agents drive the evolution of the universe; surfaces is both fully absorbed and holo- 36
222 37 are encoded on the interface. All are eigen- the dynamics of a universe entirely com- graphically encoded. Black holes are also 37
38 forms. Each agent’s internal, “bulk” states posed of agents is nothing beyond the com- ideal actors: they constantly alter the states 38
39 are experientially inaccessible to it, even bination of all of their actions. Yet from the of their environments by emitting Hawking 39
40 though each of them is on the interface of (theoretical) perspective of the entire uni- radiation. These observations and actions 40
41 and hence encodes accessible experience for verse, none of the boundaries matters. De- are classical: they can be observed by (i.e., 41
42 some agent. To see this in the simpler arena compositional equivalence allows the eras- can encode information on the interface of) 42
43 of spacetime, think of the constant experi- ing of all boundaries with no effect. From an external observer. When the situation is 43
44 ences of your own neurons (of which Cook the perspective of the whole universe, there viewed quantum-mechanically, however, 44
45 2008 provides a compelling description), all is no spacetime (indeed no classical infor- on the two sides of a black hole’s bound- 45
46 of which are inaccessible to you. mation) and nothing is happening. The uni- ary are simply quantum states, which to 46
47 « 17 » Expanding one’s (theoretical) per- verse is in a pure entangled state. That this preserve unitarity must be entangled. The 47
48 spective to the entire universe considered as fixed point exists is the physical content of correlations that implement this entangle- 48
49 a whole, however, produces not Kauffman’s the Wheeler-DeWitt equation. ment cross the boundary; they are the soft 49
50 hoped-for abduction but Dietrich’s dialetheic « 19 » The paradox posed by the “uni- quantum hairs. In Andrew Strominger’s 50
51 paradox. As described in §7 of our target versal view” is, however, deeper still. The formalism, these soft hairs are the decoher- 51
52 article, both classical and quantum physics boundary erasure allowed by decomposi- ing environment for the Hawking radiation; 52
53 allow inter-system boundaries to be moved tional equivalence erases all interfaces and the latter is detectable by us only because 53
54 or erased arbitrarily without affecting joint- hence all encoded experience. From the the soft hairs are there. The soft hairs them- 54
55 system dynamics (e.g., Zanardi 2001; Dugíc (theoretical) perspective of the entire uni- selves, however, are not detectable; they 55
column A column B column C

Constructivist Foundations vol. 12, N°3

 Second-Order Cybernetics
TOC Antonino Marcianò

column A column B column C

1 carry zero energy and hence cannot encode persistent objects located in 3D space. We mean to say that the informational redun- 1
2 classical bits on an interface. can expect both to experience the difference dancy enabled by spacetime or any other 2
3 « 22 » The interfaces of black holes, our between well-being and its absence (Peil group structure corrects errors (Q1)? As 3
4 ideal observers, are thus more complicated Kauffman 2015). What these experiences Kauffman notes, an experience per se simply 4
5 than is depicted in Figure 3 of our target are like for each individual experiencer, is what it is; there is no sense in calling it an 5
6 article. Not only do they encode classical however, remains unanswerable. It remains “error.” The errors that are corrected, in our 6
7 information; they are also a locus of quan- unanswerable, we would argue, even from a view, are errors of association between ex- 7
8 tum correlation. The former cannot happen first-person perspective. What is the experi- periences and actions. Depth perception, for 8
9 without the latter. If the encoded classical ence of green like? It is like green! Even elab- example, enables accurate grasping; disrupt- 9
10 information is the content of recallable, re- orating, saying that green is more like cyan ing depth perception introduces errors. In 10
11 portable, classical experience, the kind of than red, contributes nothing to capturing some cases, experience-action associations 11
12 experience that can be remembered or put in non-experiential terms the experience of are mediated by intervening experiences. 12
13 into a sentence, then it is natural to regard greenness. Remembering and then describ- An accurate representation of the time be- 13
14 the boundary-crossing non-classical corre- ing the greenness makes it, if anything, less tween a current sensory experience and a 14
15 lations as a kind of ineffable, non-classical immediate and vivid. Forcing experience remembered experience – as encoded in an 15
16 experience that can be neither remembered into language, even first-person language, experience of recall happening now – may 16
17 nor reported. Without this ineffable experi- distances it. be required to choose an appropriate action, 17
18 ence, recallable, reportable experience could « 25 » Holography provides a mecha- e.g., whether to hurry to avoid being late. It 18
19 not occur. nism for rendering experience classical. is errors of this sort that can decrease fitness, 19
20 « 23 » If all of the boundaries in the Beyond that, answering the “what sorts of and in extreme cases send fitness toward 20
21 universe are erased, the classical, reportable experiences?” question requires the inves- zero, stopping further input. For an organ- 21
22 experience disappears. It is, as noted ear- tigation and modeling of the particular in- ism, no action is repeated ad infinitum and 22
23 lier, epiphenomenal from a whole-universe terfaces of particular kinds of systems – e.g., no eigenform is stable forever. In a universe 23
24 perspective. The non-classical experience, particular kinds of organisms – or even where you cannot always get what you want, 24
25 however, remains. The quantum correla- particular individuals. It requires us to take you are better off having an interface that 25
26 tions that implement this non-classical ex- Werner’s questions (§14) about the structures gets you what you need. 26
27 perience constitute the universal entangled of sensory and cognitive systems seriously. 27
28 state, the fixed point of the universe’s time- Such questions inevitably lead to the field Received: 4 July 2017 28
29 less evolution. Hence Kauffman’s abduction station, the laboratory, or the clinic. It is, Accepted: 7 July 2017 29
30 can be partially recovered: the universe once again, a fair challenge to ask how and 30
31 remains filled with ineffable, non-classical even if such investigations can be fully and 31
32 experience even when all observer–system adequately described within a purely con- 32
33 boundaries have been removed. Perhaps structivist framework. We doubt it. 33
34 Kauffman’s “places of ambiguity” (§4) point « 26 » Framed in Marcianò’s terms, 34
35 to this non-classical experience as surely “what sorts of experiences?” becomes a 35
36 as do Dietrich’s dialetheia. William James’s mathematical question about the formal 36
37 (1892) “fringe” of consciousness similarly structures of model interfaces. Given an ob- 37 223
38 seems to point here. server-environment pair, for example, what 38
39 group structures characterize their interface 39
40 “What is it like?” is not one (Q3)? We have addressed this question from 40
41 question but two the reverse direction, showing that an inter- 41
42 « 24 » Kordeš (§14) introduces the tradi- face with a given group structure imposes 42
43 tional distinction between what conscious- that structure on the experienced world 43
44 ness does and what it is like, suggesting that (Hoffman, Singh & Prakash 2015 and cur- 44
45 we may address the former but can say rent work). For a finite interface and hence 45
46 nothing about the latter. We disagree, for we a finite classical experience space X, such 46
47 claim that “what is it like?” is two distinct groups are finite; hence they can at best ap- 47
48 questions. One asks what sorts of experiences proximate continuous group transforma- 48
49 might we expect a system to have, while the tions, e.g., those of the Lorentz group (Q2). 49
50 other asks what each of those experiences Whether the CA formalism can replicate 50
51 is like for each system that has it. The first the graph structures employed by physicists 51
52 of these questions can be answered, maybe while maintaining its intended interpreta- 52
53 not in all cases, but in some. We can expect tion is a topic of ongoing investigation. 53
54 bacteria, for example, to experience salti- « 27 » Kauffman raises a general question 54
55 ness and expect humans to experience time- about encoded experiences: what does it 55
column A column B column C

http://constructivist.info/12/3/200.fields
 column A column B column C
1 Combined references Cook N. D. (2008) The neuron-level phenomena Reprinted in: Foerster (2003: 169–189) 1
2 underlying cognition and consciousness: ▶︎ http://cepa.info/1637 2
3 Adams R. A., Friston K. J. & Bastos A. M. (2015) Synaptic activity and the action potential. Foerster H. von (1973) On constructing a 3
4 Active inference, predictive coding and corti- Neuroscience 153(3): 556–570. reality. In: Preiser F. E. (ed.) Environmen- 4
5 cal architecture. In: Casanova M. F. & Opris I. Csibra G. & Gergely G. (2012) Teleological tal design research. Volume II. Dowden, 5
6 (eds.) Recent advances in the modular organi- understanding of actions. In: Banaji M. R. Hutchinson & Ross, Stroudsburg PA: 35–46. 6
7 zation of the cortex. Springer, Berlin: 97–121. & Gelman S. A. (eds.) Navigating the social Reprinted in: Foerster (2003: 211–227) 7
8 Arkani-Hamed N. & Trnka J. (2014) The ampli- world: What infants, children and other spe- ▶︎ http://cepa.info/1278 8
9 tuhedron. Journal of High Energy Physics cies can teach us. Oxford University Press, Foerster H. von (1976) Objects: Tokens for 9
10 2014: 30. Oxford: 38–43. (eigen-) behaviors. ASC Cybernetics 10
11 Ashby W. R. (1956) Introduction to cybernetics. Dietrich E. & Fields C. (2015) Science generates Forum 8(3–4): 91–96. Reprinted in Foerster 11
12 Chapman and Hall, London. limit paradoxes. Axiomathes 25(4): 409–432. H. von (2003) Understanding under- 12
13 Baillargeon R. (2008) Innate ideas revisited: For Dietrich E. & Hardcastle V. (2005) Sisyphus’s standing. Springer, New York: 261–271. 13
14 a principle of persistence in infants’ physical boulder: Consciousness and the limits of the ▶︎ http://cepa.info/1270 14
15 reasoning. Perspectives on Psychological Sci- knowable. John Benjamins, New York. Foerster H. von (1979) Cybernetics of cybernet- 15
Physics Concepts in Second-Order Cybernetics

16 ence 3(1): 2–13. Dugíc M. & Jekníc-Dugíc J. (2008) What is “sys- ics. In: Krippendorf K. (ed.) Communication 16
17 Bargh J. A. & Chartrand T. L. (1999) The tem”: The information-theoretic arguments. and control. Gordon & Breach, New York: 17
18 unbearable automaticity of being. American International Journal of Theoretical Physics 5–8. Reprinted in: Foerster (2003: 283–286) 18
19 Psychologist 54(7): 462–479. 47(3): 805–813. ▶︎ http://cepa.info/1707 19
20 Bateson G. (1987) Steps to an ecology of mind. D’Ariano G. M. & Perinotti P. (2017) Quantum Foerster H. von (1981) Notes on an epistemol- 20
21 Jason Aronson, London. Originally published cellular automata and free quantum field ogy for living things. In: Foerster H. von, 21
22 in 1972. theory. Frontiers in Physics 12: 120301. Observing Systems. Intersystems Publica- 22
23 Bekenstein J. D. (1973) Black holes and entropy. Eibenberger S., Gerlich S., Arndt M., Mayor tions, Seaside CA: 258–271. 23
24 Physical Review D 7(8): 2333–2346. M. & Txen J. (2013) Matter-wave interfer- Foerster H. von (1984) On constructing a reality. 24
25 Bennett C. H. (2003) Notes on Landauer’s Prin- ence of particles selected from a molecular In: Watzlawick P. (ed.) The invented reality. 25
26 ciple, reversible computation, and Maxwell’s library with masses exceeding 10,000 amu. W. W. Norton, New York: 41–62. 26
27 Demon. Studies in the History and Philoso- Physical Chemistry and Chemical Physics Foerster H. von (2003) Understanding under- 27
28 phy of Modern Physics 34(3): 501–510. 15: 14696–14700. standing. Springer, New York. 28
29 Bilson-Thompson S. O., Markopoulou F. & Fields C. (2012) A model-theoretic interpreta- Friston K. (2010) The free-energy principle: A 29
30 Smolin L. (2007) Quantum gravity and the tion of environment-induced superselection. unified brain theory? Nature Reviews Neu- 30
31 standard model. Classical and Quantum International Journal of General Systems roscience 11: 127–138. 31
32 Gravity 24(16): 3975–3993. 41(8): 847–859. Friston K. (2013) Life as we know it. Journal of 32
33 Blackmore S. (2013) Consciousness: An introduc- Fields C. (2013) A whole box of Pandoras: Sys- the Royal Society: Interface 10: 20130475. 33
34 tion. Routledge, London. tems, boundaries and free will in quantum Friston K., Levin M., Sengupta B. & Pezzulo G. 34
35 Bohr N. (1928) The quantum postulate and the theory. Journal of Experimental and Theo- (2015) Knowing one’s place: A free-energy 35
36 recent developments of atomic theory. Nature retical Artificial Intelligence 25(3): 291–302. approach to pattern regulation. Journal of 36
224 37 121: 580–590. Fields C. (2014) Consistent quantum mechan- the Royal Society: Interface 12: 20141383. 37
38 Bohr N. (1958) Atomic physics and human ics admits no mereotopology. Axiomathes Fuchs C. (2010) QBism: The perimeter of quan- 38
39 knowledge. Wiley, New York. 24(1): 9–18. tum Bayesianism. arXiv: 1003.5209. 39
40 Bostrom N. (2003) Are we living in a computer Fields C. (2016a) Building the observer into the Fuchs C. A. & Stacey B. C. (2016) Some negative 40
41 simulation? Philosophical Quarterly 53(11): system: Toward a realistic description of remarks on operational approaches to quan- 41
42 243–255. human interaction with the world. Systems tum theory. In: Chiribella G. & Spekkens R. 42
43 Bousso R. (2002) The holographic principle. Re- 4: 32. W. (eds.) Quantum theory: Informational 43
44 views of Modern Physics 74(3): 825–874. Fields C. (2016b) Decompositional equivalence: foundations and foils. Springer, Berlin: 44
45 Chalmers D. (1995) Facing up to the problem A fundamental symmetry underlying quan- 283–305. 45
46 of consciousness. Journal of Consciousness tum theory. Axiomathes 26(3): 279–311. Geisler W. S. & Diehl R. L. (2003) A Bayesian 46
47 Studies 2(3): 200–219. Foerster H. von (1960) On self-organizing approach to the evolution of perceptual and 47
48 Clark A. & Chalmers D. (1998) The extended systems and their environments. In: cognitive systems. Cognitive Science 27(3): 48
49 mind. Analysis 58: 10–23. Yovits M. C. & Cameron S. (eds.) Self- 379–402. 49
50 Coleman S. R., Preskill J. & Wilczek F. (1992) organizing systems. Pergamon, London: Ghirardi G. C., Rimini A. & Weber T. (1986) 50
51 Quantum hair on black holes. Nuclear Physics 31–50. Reprinted in: Foerster (2003: 1–19) Unified dynamics for microscopic and 51
52 B 378(1–2): 175–246. ▶︎ http://cepa.info/1593 macroscopic systems. Physical Review D 52
53 Conway J. & Kochen S. (2006) The free will Foerster H. von (1970) Thoughts and notes on 34(2): 470–491. 53
54 theorem. Foundations of Physics 36(10): cognition. In: Gavin P. (ed.) Cognition: A Gibson J. J. (1979) The ecological approach to 54
55 1441–1473. multiple view. Spartan, New York: 25–48. visual perception. Houghton-Miffin, Boston. 55
column A column B column C

Constructivist Foundations vol. 12, N°3

 Second-Order Cybernetics
Combined references Antonino Marcianò

column A column B column C

1 Glanville R. (1982) Inside every white box ics and Human Knowing 10(3–4): 73–90. Moore E. F. (1956) Gedankenexperiments on 1
2 there are two black boxes trying to get ▶︎ http://cepa.info/1817 sequential machines. In: Shannon C. W. 2
3 out. Behavioral Science 27(1): 1–11. Kauffman L. H. (2009) Reflexivity and & McCarthy J. (eds.) Automata studies. 3
4 ▶︎ http://cepa.info/2365 eigenform: The shape of process. Con- Princeton University Press, Princeton NJ: 4
5 Glasersfeld E. von (1981) The concepts of adap- structivist Foundations 4(3): 121–137. 129–155. 5
6 tation and viability in a constructivist theory ▶︎ http://constructivist.info/4/3/121 Nagel T. (1989) The view from nowhere. Oxford 6
7 of knowledge. In: Sigel I. E., Brodzinsky D. Kauffman L. H. (2011) Eigenforms and University Press, Oxford. 7
8 M. & Golinkoff R. M. (eds.) New directions quantum physics. Cybernetics and Neumann J. von (1955) Mathematical founda- 8
9 in Piagetian theory and practice. Erlbaum, Human Knowing 18(3–4): 111–121. tions of quantum mechanics. Princeton 9
10 Hillsdale NJ: 87–95. ▶︎ http://cepa.info/1357 ▶︎ http://cepa.info/1818 University Press: Princeton NJ. German 10
11 Goodale M. A. & Milner A. D. (1992) Separate Kordeš U. (2016a) Going beyond theory. originally published in 1935. 11
12 visual pathways for perception and action. Constructivist Foundations 11(2): 375–385. Ollivier H., Poulin D. & Zurek W. H. (2004) 12
13 Trends in Neurosciences 15(1): 20–25. ▶︎ http://constructivist.info/11/2/375 Objective properties from subjective 13
14 Harshman N. L. & Ranade K. S. (2011) Observ- Kordeš U. (2016b) Where is consciousness? quantum states: Environment as a witness. 14
15 ables can be tailored to change the entangle- Constructivist Foundations 11(3): 552–554. Physical Review Letters 93: 220401. 15
16 ment of any pure state. Physical Review A Landauer R. (1961) Irreversibility and heat Ollivier H., Poulin D. & Zurek W. H. (2005) 16
17 84(1): 012303. generation in the computing process. IBM Environment as a witness: Selective prolif- 17
18 Hawking S. W. (2015) The information paradox Journal of Research and Development 5(3): eration of information and emergence of 18
19 for black holes. arXiv: 1509.01147. 183–195. objectivity in a quantum universe. Physical 19
20 Hawking S. W., Perry M. J. & Strominger A. Landauer R. (1999) Information is a physical Review A 72: 042113. 20
21 (2016) Soft hair on black holes. Physical entity. Physica A 263(1–4): 63–67. Palmer S. E. (1999) Vision science: Photons to 21
22 Review Letters 116(23): 231301. Landsman N. P. (2007) Between classical and phenomenology. MIT Press, Cambridge 22
23 Hensen B., Bernien H., Dréau A. E., Reiserer quantum. In: Butterfield J. & Earman J. (eds.) MA. 23
24 A., Kalb N., Blok M. S., Ruitenberg J., Handbook of the philosophy of science: Pastawski F., Yoshida B., Harlow D. & Preskill 24
25 Vermeulen R. F. L., Schouten R. N., Abel- Philosophy of physics. Elsevier, Amsterdam: J. (2015) Holographic quantum error- 25
26 lán C. & Amaya W. (2015) Loophole-free 417–553. correcting codes: Toy models for the bulk/ 26
27 Bell inequality violation using electron spins Levin M. A. & Wen X.-G. (2005) String-net boundary correspondence. Journal of High 27
28 separated by 1.3 kilometres. Nature 526: condensation: A physical mechanism for Energy Physics 2015: 149. 28
29 682–686. topological phases. Physical Review B 71(4): Pattee H. H. (2001) The physics of symbols: 29
30 Hickok G. & Poeppel D. (2007) The cortical 045110. Bridging the epistemic cut. Biosystems 30
31 organization of speech processing. Nature Libet B., Gleason C. A., Wright E. W. & Pearl 60(1–3): 5–21. 31
32 Reviews Neuroscience 8: 393–402. D. K. (1983) Time of conscious intention to Pearl J. (1988) Probabilistic reasoning in intel- 32
33 Hoffman D. D. & Prakash C. (2014) Objects act in relation to onset of cerebral activity ligent systems: Networks of plausible infer- 33
34 of consciousness. Frontiers in Psychology (readiness–potential) Brain 106(3): 623–642. ence. Morgan Kaufmann, San Mateo CA. 34
35 5: 577. Luo Y. & Baillargeon R. (2010) Toward a Peil Kauffman K. (2015) Emotional sentience 35
36 Hoffman D. D., Singh M. & Prakash C. (2015) mentalistic account of early psychological and the nature of phenomenal experience. 36
37 The interface theory of perception. Psycho- reasoning. Current Directions in Psychologi- Progress in Biophysics and Molecular Biol- 37 225
38 nomic Bulletin & Review 22(6): 1480–1506. cal Science 19(5): 301–307. ogy 119(3): 545–562. 38
39 Husserl E. (1982) Ideas pertaining to a pure Mandelbrot B. (1982) The fractal geometry of Penrose R. (1996) On gravity’s role in quantum 39
40 phenomenology and to a phenomenological nature. Freeman, San Francisco. state reduction. General Relativity and 40
41 philosophy. First book: General introduc- Manning A. G., Khakimov R. I., Dall R. G. & Gravitation 28(5): 581–600. 41
42 tion to a pure phenomenology. Translated Truscott A. G. (2015) Wheelers’ delayed- Pizlo Z., Li Y., Sawada T. & Steinman R. M. 42
43 by Fred Kersten. Nijhoff, The Hague. Ger- choice gedanken experiment with a single (2014) Making a machine that sees like us. 43
44 man original published in 1913. atom. Nature Physics 11: 539–542. Oxford University Press, New York. 44
45 Husserl E. (2012) Ideas: General introduction to Mark J. T., Marion B. B. & Hoffman D. D. Quine W. V. O. (1960) Word and object. MIT 45
46 pure phenomenology. Routledge, London. (2010) Natural selection and veridical Press, Cambridge MA. 46
47 German original published in 1913. perceptions. Journal of Theoretical Biology Quine W. V. O. (1970) On the reasons for the 47
48 Initials? → Misner C., Thorne K. & Wheeler 266(4): 504–515. indeterminacy of translation. Journal of 48
49 J. (1973) Gravitation. W. H. Freeman, San Marr D. (1982) Vision. Freeman, San Francisco. Philosophy 67(6): 178–183. 49
50 Francisco CA. Martin A. (2007) The representation of object Riegler A. (2012) Constructivism. In: L’Abate 50
51 James W. (1892) Psychology. University of Notre concepts in the brain. Annual Review of L. (ed.) Paradigms in theory construc- 51
52 Dame Press, Notre Dame IN. Reprinted Psychology 58: 25–45. tion. Springer, New York: 235–255. 52
53 2001 by Dover Press, Mineola NY. Maturana H. R. & Varela F. J. (1980) Autopoiesis ▶︎ http://cepa.info/446 53
54 Kauffman L. H. (2003) Eigenforms – Objects and cognition: The realization of the living. Roederer J. (2005) Information and its role in 54
55 as tokens for eigenbehaviors. Cybernet- Reidel, Boston. nature. Springer, Berlin. 55
column A column B column C

http://constructivist.info/12/3/200.fields
 column A column B column C
1 Rovelli C. (2004) Quantum gravity. Cambridge Trivers R. L. (2011) The folly of fools. Basic 1
2 University Press, Cambridge. Books, New York. 2
3 Rubino G., Rozema L. A., Feix A., Araújo M., Varela F. J., Thompson E. & Rosch E. (1991) The 3
4 Zeuner J. M., Procopio L. M., Brukner Č. & embodied mind: Cognitive science and hu- 4
5 Walther P. (2017) Experimental verifica- man experience. MIT Press, Cambridge MA. 5
6 tion of an indefinite causal order. Science Wallace D. (2008) Philosophy of quantum 6
7 Advances 3(3): E1602589. mechanics. In: Rickles D. (ed.) The Ashgate 7
8 Saini A. & Stojkovic D. (2015) Radiation from companion to contemporary philosophy of 8
9 a collapsing object is manifestly unitary. physics. Ashgate, Aldershot: 16–98. 9
10 Physical Review Letters 114: 111301. Wegner D. M. (2003) The mind’s best trick: How 10
11 Schlosshauer M. (2006) Experimental motiva- we experience conscious will. Trends in 11
12 tion and empirical consistency of minimal Cognitive Sciences 7(2): 65–69. 12
13 no-collapse quantum mechanics. Annals of Weinberg S. (2012) Collapse of the state vector. 13
14 Physics 321(1): 112–149. Physical Review A 85: 062116. 14
15 Schlosshauer M. (2007) Decoherence and the Wheeler J. A. (1990) Information, physics, 15
Physics Concepts in Second-Order Cybernetics

16 quantum to classical transition. Springer, quantum: The search for links. In: Zurek W. 16
17 Berlin. (ed.) Complexity, entropy and the physics of 17
18 Smith B. (1996) Mereotopology: A theory of information. Westview, Boulder CO: 3–28. 18
19 parts and boundaries. Data and Knowledge Wigner E. P. (1962) Remarks on the mind-body 19
20 Engineering 20(3): 287–303. question. In: Good I. J. (ed.) The Scientist 20
21 Smith J. E. & Nair R. (2005) The architecture Speculates. Basic Books, New York: 284–302. 21
22 of virtual machines. IEEE Computer 38(5): Winograd T. & Flores F. (1986) Understanding 22
23 32–38. computers and cognition: A new foundation 23
24 Spelke E. S. (1994) Initial knowledge: Six sugges- for design. Ablex Publishing, Norwood. 24
25 tions. Cognition 50(1–3): 431–445. Zahavi D. (2004) Phenomenology and the 25
26 Spencer Brown G. (1969) Laws of form. Allen & project of naturalization. Phenomenology 26
27 Unwin, London. and the cognitive sciences 3(4): 331–347. 27
28 Strominger A. (2017) Black hole information ▶︎ http://cepa.info/2375 28
29 revisited. Preprint arxiv: 1706.07143v1 Zanardi P. (2001) Virtual quantum subsystems. 29
30 (hep-th) Physical Review Letters 87(7): 077901. 30
31 Susskind L. (1995) The world as a hologram. Zeh D. (1970) On the interpretation of measure- 31
32 Journal of Mathematical Physics 36: ment in quantum theory. Foundations of 32
33 6377–6396. Physics 1(1): 69–76. 33
34 Susskind L. (2016) Computational complexity Zeh D. (1973) Toward a quantum theory of 34
35 and black hole horizons. Fortschritte Physik observation. Foundations of Physics 3(1): 35
36 64(1): 24–43. 109–116. 36
226 37 Swingle B. (2012) Entanglement renormaliza- Zurek W. H. (2003) Decoherence, einselection, 37
38 tion and holography. Physical Review D 86: and the quantum origins of the classical. 38
39 065007. Reviews of Modern Physics 75(3): 715–775. 39
40 Tarski A. (1944) The semantic conception of 40
41 truth and the foundations of semantics. 41
42 Philosophy and Phenomenological Research 42
43 4(3): 341–376. 43
44 Tegmark M. (2012) How unitary cosmology 44
45 generalizes thermodynamics and solves 45
46 the inflationary entropy problem. Physical 46
47 Review D 85: 123517. 47
48 Thompson D’A. W. (1945) On growth and form. 48
49 Cambridge University Press, Cambridge. 49
50 Originally published in 1917. 50
51 ’t Hooft G. (1993) Dimensional reduction 51
52 in quantum gravity. In: Ali A., Ellis J. & 52
53 Randjbar-Daemi S. (eds.) Salamfestschrift. 53
54 World Scientific, Singapore: 284–296. 54
55 55
column A column B column C

Constructivist Foundations vol. 12, N°3