PDF Philosophy and Computing Essays in Epistemology Philosophy of Mind Logic and Ethics 1st Edition Thomas M. Powers (Eds.) download

Download the Full Version of textbook for Fast Typing at textbookfull.
com
Philosophy and Computing Essays in Epistemology

Philosophy of Mind Logic and Ethics 1st Edition
Thomas M. Powers (Eds.)
https://textbookfull.com/product/philosophy-and-computing-
essays-in-epistemology-philosophy-of-mind-logic-and-
ethics-1st-edition-thomas-m-powers-eds/
OR CLICK BUTTON
DOWNLOAD NOW
Download More textbook Instantly Today - Get Yours Now at textbookfull.com

Recommended digital products (PDF, EPUB, MOBI) that
you can download immediately if you are interested.
Philosophy in Educational Research Epistemology Ethics

Politics and Quality 1st Edition David Bridges (Auth.)
https://textbookfull.com/product/philosophy-in-educational-research-
epistemology-ethics-politics-and-quality-1st-edition-david-bridges-
auth/
textboxfull.com
Beyond Faith and Rationality Essays on Logic Religion and

Philosophy Ricardo Sousa Silvestre
https://textbookfull.com/product/beyond-faith-and-rationality-essays-
on-logic-religion-and-philosophy-ricardo-sousa-silvestre/
textboxfull.com
The Philosophy of Piers Plowman: The Ethics and

Epistemology of Love in Late Medieval Thought 1st Edition
David Strong (Auth.)
https://textbookfull.com/product/the-philosophy-of-piers-plowman-the-
ethics-and-epistemology-of-love-in-late-medieval-thought-1st-edition-
david-strong-auth/
textboxfull.com
Philosophy of Mind in the Twentieth and Twenty First

Centuries The History of the Philosophy of Mind 1 edition
Edition Amy Kind
https://textbookfull.com/product/philosophy-of-mind-in-the-twentieth-
and-twenty-first-centuries-the-history-of-the-philosophy-of-
mind-1-edition-edition-amy-kind/
textboxfull.com
Essays in the Philosophy of Chemistry 1st Edition Fisher
https://textbookfull.com/product/essays-in-the-philosophy-of-
chemistry-1st-edition-fisher/
textboxfull.com
Ethics Beyond the Limits New Essays on Bernard Williams

Ethics and the Limits of Philosophy Sophie Grace Chappell
https://textbookfull.com/product/ethics-beyond-the-limits-new-essays-
on-bernard-williams-ethics-and-the-limits-of-philosophy-sophie-grace-
chappell/
textboxfull.com
Grammar, Philosophy, and Logic 1st Edition Bruce Silver

(Auth.)
https://textbookfull.com/product/grammar-philosophy-and-logic-1st-
edition-bruce-silver-auth/
textboxfull.com
Implicit bias and philosophy. Volume 1, Metaphysics and

epistemology 1st Edition Brownstein
https://textbookfull.com/product/implicit-bias-and-philosophy-
volume-1-metaphysics-and-epistemology-1st-edition-brownstein/
textboxfull.com
Logic and Philosophy of Time Themes from Prior 1st Edition

Hasle
https://textbookfull.com/product/logic-and-philosophy-of-time-themes-
from-prior-1st-edition-hasle/
textboxfull.com
Philosophical Studies Series
Thomas M. Powers Editor
Philosophy
and
Computing
Essays in Epistemology, Philosophy of
Mind, Logic, and Ethics
Volume 128
Editor-in-Chief
Luciano Floridi, University of Oxford, Oxford Internet Institute, United Kingdom
Mariarosaria Taddeo, University of Oxford, Oxford Internet Institute, United Kingdom
Executive Editorial Board

Patrick Allo, Vrije Universiteit Brussel, Belgium
Massimo Durante, Università degli Studi di Torino, Italy
Phyllis Illari, University College London, United Kingdom
Shannon Vallor, Santa Clara University
Board of Consulting Editors

Lynne Rudder Baker, University of Massachusetts at Amherst
Stewart Cohen, Arizona State University, Tempe
Radu Bogdan, Tulane University
Marian David, University of Notre Dame
John M. Fischer, University of California at Riverside
Keith Lehrer, University of Arizona, Tucson
Denise Meyerson, Macquarie University
François Recanati, Institut Jean-Nicod, EHESS, Paris
Mark Sainsbury, University of Texas at Austin
Barry Smith, State University of New York at Buffalo
Nicholas D. Smith, Lewis & Clark College
Linda Zagzebski, University of Oklahoma
More information about this series at http://www.springer.com/series/6459
Thomas M. Powers
Editor
Philosophy and Computing

Essays in Epistemology, Philosophy of Mind,
Logic, and Ethics
123
Editor
Thomas M. Powers
Department of Philosophy
University of Delaware
Newark, DE, USA
ISSN 0921-8599 ISSN 2542-8349 (electronic)

ISBN 978-3-319-61042-9 ISBN 978-3-319-61043-6 (eBook)
DOI 10.1007/978-3-319-61043-6
Library of Congress Control Number: 2017952948
© Springer International Publishing AG 2017

This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of
the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation,
broadcasting, reproduction on microfilms or in any other physical way, and transmission or information
storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology
now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication
does not imply, even in the absence of a specific statement, that such names are exempt from the relevant
protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this book
are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or
the editors give a warranty, express or implied, with respect to the material contained herein or for any
errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional
claims in published maps and institutional affiliations.
Printed on acid-free paper
This Springer imprint is published by Springer Nature

The registered company is Springer International Publishing AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
Contents
1 Introduction: Intersecting Traditions in the Philosophy

of Computing. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
Thomas M. Powers
2 Levels of Computational Explanation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
Michael Rescorla
3 On the Relation of Computing to the World. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
William J. Rapaport
4 Cognitive Computation sans Representation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
Paul Schweizer
5 Software Error as a Limit to Inquiry for Finite Agents:
Challenges for the Post-human Scientist . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
John F. Symons and Jack K. Horner
6 The Singularity Business. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
Selmer Bringsjord and Alexander Bringsjord
7 Artificial Moral Cognition: Moral Functionalism
and Autonomous Moral Agency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
Don Howard and Ioan Muntean
8 AI and the Automation of Wisdom . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161
Shannon Vallor
9 An Analysis of Machine Ethics from the Perspective of Autonomy . . . 179
Mario Verdicchio
10 Beyond Informed Consent—Investigating Ethical Justifications
for Disclosing, Donating or Sharing Personal Data in Research . . . . . . 193
Markus Christen, Josep Domingo-Ferrer, Dominik Herrmann,
and Jeroen van den Hoven
v
vi Contents
11 Big Data, Digital Traces and the Metaphysics of the Self . . . . . . . . . . . . . . 209
Soraj Hongladarom
12 Why Big Data Needs the Virtues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
Frances S. Grodzinsky
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 235
Chapter 1
Introduction: Intersecting Traditions
in the Philosophy of Computing
Thomas M. Powers
Keywords Philosophy of computing • Cognition • Representation • Autonomy

• Big Data • Machine ethics
This volume consists of selected papers from the 2015 joint international
conference—the first-ever meeting of the Computer Ethics-Philosophical Enquiry
conference series of the International Society for Ethics and Information
Technology, and the International Association for Computing and Philosophy—held
at the University of Delaware from June 22–25 of 2015. The organizing themes of
the conference are well represented in the volume. They include theoretical topics at
the intersection of computing and philosophy, including essays that explore current
issues in epistemology, philosophy of mind, logic, and philosophy of science,
and also normative topics on matters of ethical, social, economic, and political
import. All of the essays provide views of their subject matter through the lens of
computation.
Two general types of question motivate the foregoing essays. First, how has
computation changed philosophical inquiry into the nature of mind and cognition?
Second, how we can come to terms with the ethical and political aspects of the
computer and information-technology revolution? It is worth noting that these
questions—though they have lingered on the surface and beneath many philosoph-
ical discussions for decades (and in some cases, for centuries)—are given a new
treatment by the authors precisely because of recent developments in the science
and technology of computation. Almost all philosophers know the general landscape
of these questions well—What is the nature of explanation? What is thought?
How does language represent the world? What is it to act in such a way as to be
responsible? Formerly, answers to these questions placed humans at the center of
such inquiries; philosophy was concerned with explanations given by humans and
for humans. We considered—as though it were tautological—only the possibility of
human (natural) languages, learning, thought, agency, and the like. But philosophy
cannot ignore the fact that computational machines are capable of joining us at the
T.M. Powers ()

Department of Philosophy, University of Delaware, Newark, DE, USA
e-mail: [email protected]
© Springer International Publishing AG 2017 1

T.M. Powers (ed.), Philosophy and Computing, Philosophical Studies Series 128,
DOI 10.1007/978-3-319-61043-6_1
2 T.M. Powers
center of this inquiry, and thus we can now ponder machine language, learning,
thought, agency, and other quite revolutionary concepts.
The impetus for the new treatments in these essays comes from startling
developments in contemporary technologies which allow pervasive information
gathering and analyses far beyond human capacities—the era of Big Data—and new
scientific insights into our own cognition. We are also beginning to see sophisticated,
autonomous, and deadly functionality in machines, and there is even talk of the
possibility of a “post-human” society due to the convergence of genetics and
genomics with cognitive neuroscience and information technologies. So indeed, we
are entering uncharted territory for science, ethics, politics, and civilization.
That philosophy would once again be forced to reconstitute itself because of
developments in science and technology would have been well appreciated by
Descartes, through his study of microscopy, as well as by Kant’s grounding of a crit-
ical metaphysics to acknowledge the contributions of Newton’s mechanics. Indeed,
the essays in this volume also fall in the tradition of philosophy reconstituting itself,
as described in Floridi’s “fourth revolution.”
The virtue of a collection that ranges over philosophical questions, such as this
one does, lies in the prospects for a more integrated understanding of issues. These
are early days in the partnership between philosophy and information technology,
and many foundational issues are still being sorted out. It is to be expected that many
of the tools of philosophy will have to be deployed to establish this foundation, and
this volume admirably showcases those tools in the hands of scholars. Here briefly
is what the reader can expect.
In Michael Rescorla’s essay “Levels of Computational Explanation” he analyzes
three levels in describing computing systems (representational, syntactic, and
physical) and argues that syntactic description is key because it mediates between
the representational description and physical construction of artificial systems.
However, while this three-level view works well for artificial computing systems,
it is inadequate for natural computing systems (e.g., humans) on his view. By
focusing on explanatory practice in cognitive science, he brings to the foreground
the pragmatic advantages of syntactic description for natural computing systems.
In “On the Relation of Computing to the World” William J. Rapaport surveys a
multitude of ways of construing said relation. One would be hard-pressed to find
a stone unturned here: questions of semantic and syntactic computation (and even
syntactic semantics), the nature and teleology of algorithms, program verification
and the limits of computation, symbols and meanings in computation, and inputs
and outputs in a Turing machine are all considered.
Paul Schweizer’s “Cognitive Computation sans Representation” addresses what
he considers to be the deep incompatibility between the computational and the
representational theories of mind (CTM vs. RTM). Attempts to join them, he argues,
are destined to fail because the unique representational content claimed by RTM is
superfluous to the formal procedures of CTM. In a computing mechanism, after
syntactic encoding “it’s pure syntax all the way down to the level of physical
implementation.”
1 Introduction: Intersecting Traditions in the Philosophy of Computing 3
In “Software Error as a Limit to Inquiry for Finite Agents: Challenges for the
Post-human Scientist” John F. Symons and Jack K. Horner revisit C.S. Peirce’s
intuition about truth being the output of inquiry in the limit by considering the
limits for the future of software-driven science. They use their previously-developed
concept of a finite post-human agent, constrained “only by mathematics and the
temporal finiteness of the universe,” to argue that greater use of software or other
rule-governed procedures will lead to decreased ability to control for error in
scientific inquiry.
Concerns about a post-human world are also at the center of Selmer and
Alexander Bringsjord’s “The Singularity Business: Toward a Realistic, Fine-grained
Economics for an AI-Infused World.” Here they are interested literally in business
and economic questions that will arise in a future with “high-but-sub-human-level
artificial intelligence.” They investigate the implications of an artificial intelligence
that is impressive and useful, from the standpoint of contemporary science and
engineering, but falls well short of the awareness or self-consciousness promised
by The Singularity. They argue that even this “minimal” artificial intelligence will
have tremendous labor market and social implications, arising from the tasks to
which it can be set.
In “Artificial Moral Cognition: Moral Functionalism and Autonomous Moral
Agency” by Don Howard and Ioan Muntean, we find a model of an Artificial
Autonomous Moral Agent (AAMA) that will engage in moral cognition by learning
moral-behavioral patterns from data. Unlike a rule-following autonomous agent,
their AAMA will be based on “soft computing” methods of neural networks and
evolutionary computation. They conceive of the resulting agent as an example of
virtue ethics for machines, having a certain level of autonomy and complexity, and
being capable of particularistic moral judgments. They apply the concept of the
AAMA to Hardin’s metaphor of the “lifeboat” in the ethics of choice with limited
resources.
Similarly, Shannon Vallor takes a virtue ethics approach to questions of machine
and robotic ethics in her “AI and the Automation of Wisdom.” Like the Bringsjords,
Vallor is concerned with economic, political, and technological implications—here,
primarily, to the future development of human wisdom in a world that may seem
to eschew it. Knowledge and skills in the coming workforce will change to meet
societal needs, as they always have, but more troubling is the threat posed by an
“algorithmic displacement of human wisdom.” Drawing on sources from Aristotle
to the present, she forewarns of a weakened ability to rise to our environmental,
civic, and political responsibilities.
Mario Verdicchio presents “An Analysis of Machine Ethics from the Perspective
of Autonomy” in order to urge a return to what he calls classic ethics to guide
researchers in machine ethics. He rejects the call for a new discipline of ethics for
machines—one that would focus on embedding ethical principles into programs. He
argues instead that industry-driven standards for robotic safety are sufficient, and
that nothing in robotics presents a fundamental challenge to the ethics of design;
rather, new machine capabilities show us why we ought to focus on the traditional
(classic) ethics that guide human choices. While he acknowledges that cutting-edge
4 T.M. Powers
robots may have a higher degree of autonomy than those in the past, he does not
think that such autonomy is sufficient to require a new ethics discipline.
We turn to questions of research ethics in the era of Big Data with “Beyond
Informed Consent: Investigating Ethical Justifications for Disclosing, Donating or
Sharing Personal Data in Research” by Markus Christen, Josep Domingo-Ferrer,
Dominik Herrmann, & Jeroen van den Hoven. In this essay the authors consider
how the modern digital research ecosystem challenges notions of informed consent,
control of personal data, and protection of research subjects from unintended effects
of research in a rapidly changing social and scientific landscape. They develop
arguments around three core values—autonomy, fairness and responsibility—to
show how an active community of research participants can be educated through
and involved in research over time. Such a community would enable user-centric
management of personal data, including generation, publication, control, exploita-
tion, and self-protection.
Soraj Hongladarom turns to ontological concerns in “Big Data, Digital Traces
and the Metaphysics of the Self.” This investigation begins with a conception of
self and identity that depends on one’s Internet activity. Identity is constituted by
“how one leaves her traces digitally on the Internet.” This view borrows from
the well-known “extended mind” thesis of Chalmers and Clark and issues in
a conception of the (digitally) extended self. Hongladarom suggests that these
traces—the distributed parts of the ontologically-whole self—nonetheless belong
to the owner. Thus, they deserve protection and generate strong claims of privacy
and respect for individuals.
In the final essay of this volume, “Why Big Data Needs the Virtues” by Frances
S. Grodzinsky, we encounter an analysis of Big Data and its value, with an argument
on how it can be harnessed for good. Grodzinsky starts with an account of Big Data’s
volume, velocity, variety, and veracity. She goes on to critique claims that statistical
correlations in Big Data are free of theory, ready to gleaned from data sets. Turning
to the ethics of the “Big Data Scientist,” she sketches a virtuous epistemic agent
who incorporates both virtue ethics and virtue epistemology. Such an agent will be
well placed, she believes, to act responsibly to use Big Data for socially-beneficial
ends.
Through analyses of the foregoing issues, the philosophical work in these
chapters promises to clarify important questions and help develop new lines of
research. It is hoped that readers will find much of value in these intersecting
traditions in philosophy and computing.
Chapter 2
Levels of Computational Explanation
Michael Rescorla
Abstract It is widely agreed that one can fruitfully describe a computing system
at various levels. Discussion typically centers on three levels: the representational
level, the syntactic level, and the hardware level. I will argue that the three-
level picture works well for artificial computing systems (i.e. computing systems
designed and built by intelligent agents) but less well for natural computing systems
(i.e. computing systems that arise in nature without design or construction by
intelligent agents). Philosophers and cognitive scientists have been too hasty to
extrapolate lessons drawn from artificial computation to the much different case
of natural computation.
Keywords Levels of explanation • Representation • Syntax • The computational

theory of mind • Intentionality • Functionalism • Abstraction • Bayesianism
2.1 Representation, Syntax, and Hardware
It is widely agreed that one can fruitfully describe a computing system at various
levels. Discussion typically centers on three levels that I will call the represen-
tational level, the syntactic level, and the hardware level. To illustrate, consider
a computer programmed to perform elementary arithmetical operations such as
addition, multiplication, and division:
– At the representational level, we individuate computational states through their
representational properties. For instance, we might say that our computer divides
the number 2 into the number 5 to yield remainder 1. This description implicitly
presupposes that the computer’s states represent specific numbers.
– At the syntactic level, we individuate computational states non-representationally.
We describe our computer as manipulating numerals, rather than performing
arithmetical operations over numbers. For example, we might say that the
computer performs certain syntactic operations over the numerals “2” and “5”
M. Rescorla ()
Department of Philosophy, University of California, Los Angeles, CA, USA
e-mail: [email protected]
© Springer International Publishing AG 2017 5

T.M. Powers (ed.), Philosophy and Computing, Philosophical Studies Series 128,
DOI 10.1007/978-3-319-61043-6_2
6 M. Rescorla
and then outputs the numeral “1.” When offering this description, we do not
presuppose that the computer’s states represent numbers.
– At the hardware level, we describe the physical realization of computational
states. We specify our computer’s components, how those components are
assembled, and how the computer’s physical state evolves according to well-
understood physical laws.
A three-level picture along these lines figures prominently in many philosophical
and scientific discussions (Chalmers 2011, 2012; Fodor 1981, 1987, 1994, 2008;
Haugeland 1985; Pylyshyn 1984).
I will argue that the three-level picture works well for artificial computing
systems (i.e. computing systems designed and built by intelligent agents) but
less well for natural computing systems (i.e. computing systems that arise in
nature without design or construction by intelligent agents). Philosophers and
cognitive scientists have been too hasty to extrapolate lessons drawn from artificial
computation to the much different case of natural computation. I discuss artificial
computation in Sects. 2 and 3 and natural computation in Sect. 4. I compare the two
cases in Sect. 5.
2.2 Representational Description of Artificial Computation
Hardware description figures indispensably within computing practice. Ultimately,

we must describe the materials from which a machine is to be built, the way those
materials are to be combined, the intended physical evolution of the machine, and
so on. Only then can we build the machine. A good hardware description serves as
a blueprint, specifying how to construct a physical system with desired properties.
Suitable hardware description is also needed for various modifications or repairs we
might make. These points are evident, so I will not discuss them further.
I focus on the representational and syntactic levels. I will argue that represen-
tational description illuminates a wide range of artificial computations (Sects. 2.1
and 2.2). I will then argue that syntactic description plays a key role in mediating
between representational description and physical construction of artificial systems
(Sect. 3).
2.2.1 Representation Elucidated
Researchers across philosophy, computer science (CS), and cognitive science use
the phrase “representation” in various ways. Following common philosophical
usage (e.g. Burge 2010, p. 9), I tie representation to veridicality-conditions. To
illustrate:
2 Levels of Computational Explanation 7
– Beliefs are evaluable as true or false. My belief that Barack Obama is president
is true if Barack Obama is president, false if he is not.
– Declarative sentences (e.g. “Barack Obama is president”) as uttered in specific
conversational contexts are likewise evaluable as true or false.
– Perceptual states are evaluable as accurate or inaccurate. A perceptual state that
represents presence of a red sphere is accurate only if a red sphere is before me.
– Intentions are evaluable as fulfilled or thwarted. My intention to eat chocolate is
fulfilled if I eat chocolate, thwarted if I do not eat chocolate.
Truth-conditions, accuracy-conditions, and fulfillment-conditions are species of
veridicality-conditions. Complex representations decompose into parts whose rep-
resentational properties contribute to veridicality-conditions. For example, the
truth-condition of “John loves Mary” is determined by the denotation of “John,” the
denotation of “Mary,” and the satisfaction-condition of “loves.” Representational
description invokes veridicality-conditions or representational properties that con-
tribute to veridicality-conditions.
I distinguish two ways that a system may come to have representational
properties: it may have representational properties at least partly by virtue of its
own activity; or it may have representational properties entirely because some
other system has imposed those properties upon it. For example, the human mind
has representational properties at least partly due to its own activity. In contrast,
words in a book represent entirely by virtue of their connection to our linguistic
conventions. The book does not contribute to representational properties of its
component words.
Philosophers commonly evoke this distinction using the labels original ver-
sus derived intentionality (Haugeland 1985) or intrinsic versus observer relative
meanings (Searle 1980). To my ear, these labels suggest that the main con-
trast concerns whether a system is solely responsible for generating its own
representational properties. Yet Burge (1982) and Putnam (1975) have argued
convincingly that the external physical and social environment plays a large role
in determining representational properties of mental states, so that not even the
mind is solely responsible for generating its own representational properties. I
prefer the labels indigenous versus inherited, which seem to me to carry fewer
misleading connotations. Representational properties of human mental states are
indigenous, because human mental activity plays at least some role in generating
representational properties of mental states. Representational properties of words
in a book are inherited, because the book plays no role in generating those
properties.
Are representational properties of artificial computing systems inherited or
indigenous? For the artificial computing systems employed in our own society, the
answer is usually “inherited.” For example, a simple pocket calculator represents
numbers entirely by virtue of our linguistic conventions regarding numerals. A sim-
ilar diagnosis applies to many far more sophisticated systems. Some philosophers
maintain that artificial computing machines in principle cannot have indigenous
representational properties (Searle 1980). I think that this position is implausible
8 M. Rescorla
and that existing arguments for it are flawed. I see no reason why a sufficiently
sophisticated robot could not confer representational properties upon its own inter-
nal states. We could equip the robot with sensors or motor organs, so that it causally
interacts with the external world in a suitably sophisticated way. So equipped, I
see no reason why the robot could not achieve indigenous representation of its
external environment. Whether any actual existing artificial computing systems have
indigenous representational properties is a trickier question that I set aside.
2.2.2 The Value of Representational Description
To what extent do we illuminate an artificial computing system by citing its

representational properties (whether those properties are inherited or indigenous)?
We often want to compute over a non-linguistic domain. Consider the Euclidean
algorithm for computing greatest common divisors (gcds). The algorithm features
arithmetical operations over natural numbers (Rescorla 2013a). Numbers cannot
be directly instantiated inside a computing system. Rather, the computing system
must instantiate numerals that denote natural numbers. More generally, a system
can compute over non-linguistic items only if the system represents those items
(Rescorla 2015c). When we describe a machine as computing over a non-linguistic
domain, we presuppose (at least implicitly) that the machine’s states represent
elements of the domain.
Non-linguistic domains studied within computability theory typically contain
mathematical entities (e.g. natural numbers). Real-world applications just as often
involve computation over a non-linguistic, non-mathematical domain. We might
want a smartphone that computes the fastest route from one location to another;
or a library catalogue system that allows users to recall library books; or a robot that
estimates current position based on sonar and odometer readings; or a program that
retrieves an individual’s city of birth from a database; and so on. Each computation
is defined at least partly over a non-linguistic, non-mathematical domain. To
understand the computation, we must describe it as representing locations, books,
people, cities, and so on. Representational description helps us articulate why we
built the machine and what function it serves in our broader practices.
Some philosophers assert or intimate that representational description is rela-
tively unimportant to computing practice (Chalmers 2011; Piccinini 2008, 2009).
Why not instead employ syntactic descriptions? Rather than say that a machine
computes the gcd of two numbers, why not say that it executes a syntactic operation
over numerals? Rather than say that a machine retrieves an individual’s city of
birth, why not say that the machine executes appropriate operations over names?
Doesn’t the representational level become superfluous once we isolate an underlying
syntactic description?
I find such proposals jarringly divergent from actual practice within
computability theory, CS, robotics, industry, and everyday life. In all these spheres,
we are often primarily concerned with representational aspects of computation. The

representational level is not superfluous, because it captures the primary purpose
served by underlying syntactic operations. Representational description is needed to
articulate the main reason why we built the machine in the first place: computation
over a non-linguistic domain.
Accordingly, representational relations between computing systems and repre-
sented domains figure crucially within scientific inquiry. Three examples:
– Computability theory studies computational properties of notations for various
non-linguistic domains (Rescorla 2015c). Different notations embody different
ways of representing the domain. For instance, computable analysis studies
which operations over real numbers are computable relative to decimal notation
and which are computable relative to alternative notations (Weihrauch 2000).
– Computer science offers rigorous computational models that describe computa-
tions in representational terms (Rescorla 2013a, 2014c). For example, one can
codify the Euclidean algorithm as a LISP program or as a register machine
program (Abelson et al. 1996, p. 49, p. 497). The resulting programs describe
computation over natural numbers, just as the pre-theoretic Euclidean algorithm
does. So the programs individuate computational states partly through their
representational properties.
– Probabilistic robotics delineates Bayesian algorithms describing how a robot
navigates through its environment (Thrun et al. 2005). These algorithms presup-
pose that the robot maintains an internal map that represents spatial aspects of the
environment (Rescorla 2009). The algorithms dictate how to update probabilities
over maps in light of sensor measurements and odometer readings.
We could not retain computability theory, computer science, or probabilistic
robotics in anything resembling their current forms if we were to jettison repre-
sentational description in favor of syntactic description.
Representational description usefully characterizes not just inputs and outputs to
computation but also internal states. Suppose we describe a machine as executing
the Euclidean algorithm. In doing so, we describe the machine as repeatedly
executing the division operation and hence as representing a series of numbers
intermediate between input and output (Rescorla 2013a). Or suppose we describe
a robot as updating probabilities over maps. The robot’s updates involve internal
representational states—probability assignments to maps that represent the envi-
ronment. In these examples, and many others, we want our machine to transit
appropriately between internal computational states with suitable representational
properties.
I do not say that representational description illuminates all artificial computing
systems. I say only that it illuminates many artificial computing systems that figure
prominently in science and everyday life.
10 M. Rescorla
2.3 Syntactic Description of Artificial Computation
There has been considerable philosophical discussion surrounding the notion of

“syntax.” I will assume that syntactic description has at least two properties:
– Syntactic description is non-representational, i.e. it does not individuate compu-
tational states through their representational properties. If we say that a computer
stores the numeral “13” in some memory register, then we have offered a non-
representational description, because the numeral “13” might have had different
representational properties (or no representational properties at all) depending on
the surrounding computational, physical, or social environment.
– Syntactic description is multiply realizable in Putnam’s (1975) sense: physical
systems with wildly heterogeneous physical properties may satisfy a given
syntactic description. Because syntactic description is multiply realizable, it is
much more abstract than hardware description.
Syntactic descriptions with these two properties figure prominently in computing
practice. Consider a Turing machine that manipulates stroke marks on a machine
tape; or a system of logic gates that manipulate “0”s and “1”s; or a desktop
computer that compiles a high-level programming language into machine code.
In each case, we can describe how the machine manipulates syntactic items while
ignoring representational and physical details.
Philosophers commonly gloss syntax in functionalist terms (Chalmers 2011;
Field 2001, pp. 56–57; Fodor 1994, pp. 108–110; Stich 1983, pp. 149–151):
syntactic states are individuated through their characteristic relations to one another
and to computational inputs and outputs. Chalmers develops the functionalist
conception using the notion of causal topology: “the pattern of interaction among
parts of the system, abstracted away from the make-up of individual parts and from
the way the causal connections are implemented” (Chalmers 2011, p. 337). On
Chalmers’s view, syntactic description specifies a causal topology. It thereby con-
strains abstract causal structure but not physical details. A variant functionalist view
allows syntactic description to constrain both abstract causal structure and physical
aspects of inputs and outputs (e.g. geometric shapes of a desktop computer’s inputs
and outputs; physical properties of a robot’s sensor inputs and motor outputs).
In everyday computing practice, we are often primarily concerned with compu-
tation over syntactic entities. We describe a computation by decomposing it into
elementary syntactic operations (e.g. moving a word from one memory register
to another) that transform syntactic inputs into syntactic outputs. For example, a
compiler carries syntactic inputs (source code in a high-level programming lan-
guage) into syntactic outputs (machine language code) through a series of syntactic
manipulations. Typically, an artificial computing system falls under multiple levels
of syntactic description. Syntactic description of a desktop computer is organized in
a hierarchy, ranging from logic gates to machine code to assembly code to a high-
level programming language. As we ascend the hierarchy, we describe progressively
more abstract aspects of syntactic processing.
Even when computer scientists are most fundamentally interested in

representational aspects of computation, syntactic description plays a pivotal role.
Suppose we want to build a machine that executes some computation as described
in representational terms. Representational description, taken on its own, does not
specify how to build such a machine. Even if we know how we want our machine
to transit between representational states, we may have little idea how to build
a machine that so transits. As Chalmers (2012, p. 245) puts it, “[o]ne cannot go
straight from representational explanation to building a mechanism; one has some
hard working to do in figuring out the right mechanism.” How do we ensure that
our machine transits as desired between representational states? How do we build
a machine that reliably transits from a computational state that bears some relation
to the represented domain to a computational state that bears some other desired
relation to the represented domain? For example, suppose we want a machine that
executes the Euclidean algorithm. How do we ensure that our machine divides
numbers into one another as the algorithm requires? Representational description
by itself does not supply anything like a workable blueprint for a physical machine.
Turing’s (1936) brilliant solution: supplement representational description with
syntactic description. To build a machine that transits appropriately between rep-
resentational states, we build a machine that manipulates syntactic items. Suitable
manipulation of syntactic items endowed with suitable representational properties
ensures that the machine satisfies our desired representational description. To build
a machine that executes the Euclidean algorithm, we isolate an algorithm for
manipulating numerals. Assuming that the numerals denote suitable numbers, the
machine thereby computes gcds.
Syntactic description carries us much closer than representational description to
a workable blueprint for a physical machine. In principle, we know how to build
a machine that executes iterated elementary syntactic operations over syntactic
items. This is especially true for low-level syntactic descriptions, such as logic
gate descriptions or machine language descriptions. It is also true for more abstract
syntactic descriptions, such as a LISP program that specifies manipulation of list
structures. Syntactic description helps us design and construct machines in a way
that representational description does not.
Why exactly does syntactic description carry us closer than representational
description to a workable blueprint? Because we do not know helpful sufficient
conditions for a machine to instantiate desired representational properties:
– Indigenous representational properties typically depend upon complex causal
interactions between the physical system and its surrounding environment—
causal interactions that we are currently unable to specify in an informative way.
Even when the represented domain is mathematical, we do not have a good
theory describing what it takes for the system to bear appropriate relations to
the represented domain. We have nothing like a useful blueprint for ensuring that
a machine has suitable indigenous representational properties.
– Inherited representational properties may initially seem less problematic, since
it is easy enough to stipulate that a machine’s internal states have certain
12 M. Rescorla
representational properties. However, we cannot ensure that other users of some

machine will make the same stipulations. Once we release a machine into the
wild, we have little control over which representational properties other people
bestow upon it. Thus, we are not able to provide anything like a useful blueprint
for ensuring that a machine has suitable inherited representational properties.
Syntactic description avoids these problems. By focusing solely on “intrinsic”
aspects of computation, without seeking to ensure that computational states bear
appropriate relations to represented entities, syntactic description carries us much
closer to a workable blueprint for a physical system.
Hardware description likewise supplies a workable blueprint. As Chalmers
(2012) emphasizes, though, it includes numerous details that are irrelevant for many
purposes. When designing or modifying a computing machine, we often do not care
about the exact physical substrate that implements, say, memory registers. We would
like a workable blueprint that prescinds from irrelevant hardware details. Syntactic
description fulfills this desideratum. As Chalmers (2012, p. 245) puts it, syntactic
description “yields a sweet spot of being detailed enough that a fully specified
mechanism is provided, while at the same time providing the minimal level of detail
needed for such a mechanism.”
Chalmers’s analysis illuminates why syntactic description figures so centrally
within computing practice. Even when we are mainly concerned with representa-
tional aspects of computation, syntactic description helps us build a physical system
that transits appropriately between representational states. Syntactic description
helps because it is non-representational (so that it furnishes a workable blueprint)
and multiply realizable (so that it suppresses irrelevant hardware details).
Chalmers’s analysis is closely related to abstraction, a common technique in
computer science (Abelson et al. 1996, pp. 83–89). Abstraction is suppression of
low-level implementation detail. For example, one might model manipulation of
list structures without specifying how list structures are implemented by memory
registers. Abstraction has several virtues:
– Abstraction helps us manage the enormous complexity of typical computing
systems. Designing and modifying complex systems is much easier when we
ignore details that do not bear upon our current design goals.
– Abstraction increases flexibility, allowing us to remain non-committal about how
exactly we will implement our high-level description. Flexibility is important if
we are not sure which low-level implementation is best, or if we want to permit
different implementation details at some future date.
The advantages of syntactic description over hardware description are a special
case of the general pressure toward abstraction. Good computer design manages
complexity and promotes flexibility by suppressing irrelevant hardware details
whenever possible.
I conclude that syntactic description advances our pragmatic computing goals
in a distinctive way that representational description and hardware description
do not. Syntactic description helps us design and build physical machines that
implement representationally-specified computations. It plays a crucial role in

mediating between representational description and physical construction of arti-
ficial computing machines.1
2.4 Natural Computing Systems
By a “natural system,” I mean one that arises in nature without design or oversight
by intelligent agents. Whether a system counts as “natural” is a matter of its etiology,
not its material constitution. A computing system constructed by humans from DNA
or other biochemical material is not “natural,” because it is an artifact. A silicon-
based creature that evolved through natural selection on another planet counts as
“natural,” even though it is not constructed from terrestrial biochemical materials.
According to the computational theory of mind (CTM), the mind is a computing
system. Classical CTM holds that the mind executes computations similar to those
executed by Turing machines (Fodor 1975, 1987, 1994, 2008; Gallistel and King
2009; Putnam 1975; Pylyshyn 1984). Connectionist CTM models mental activity
using neural networks (Horgan and Tienson 1996; Ramsey 2007; Rumelhart et al.
1986). Both classical and connectionist CTM trace back to seminal work of McCul-
loch and Pitts (1943). In (Rescorla 2015b), I surveyed classical, connectionist,
and other versions of CTM. For present purposes, I do not assume any particular
version of CTM. I simply assume that the mind in some sense computes. Under
that assumption, it makes sense to talk about “natural computing systems.” We may
therefore ask how Sect. 1’s levels of description apply to natural computation—
specifically, mental computation.2
Hardware description is vital to the study of mental computation. Ultimately, we
want to know how neural tissue physically realizes mental computations. Everyone
agrees that a complete cognitive science will include detailed hardware descriptions
that characterize how neural processes implement mental activity. Unfortunately,
satisfactory hardware descriptions are not yet available. Although we know quite a
1
When representational properties are inherited rather than indigenous, syntactic description offers
further advantages over representational description. I argue in (Rescorla 2014b) that inherited
representational properties of computational states are causally irrelevant: one can freely vary
inherited representational properties without altering the underlying syntactic manipulations, so
representational properties do not make a difference to the computation. Representational descrip-
tion does not furnish genuinely causal explanations of a system whose representational properties
are all inherited. No such analysis applies to a computational system whose representational
properties are indigenous. In that case, I claim, representational properties can be causally relevant
(Rescorla 2014b).
2
For purposes of this paper, “mental computation” indicates computation by a natural system with
a mind. I leave open the possibility that an artificial system (such as a sophisticated robot) might
also have a mind.
14 M. Rescorla
lot about the brain, we still do not know how exactly neural processing physically
realizes basic mental activities such as perception, motor control, navigation,
reasoning, decision-making, and so on.
What about representational and syntactic description? Will these also figure in
any complete cognitive science? I discuss representation in Sect. 4.1 and syntax in
Sect. 4.2.
2.4.1 Representational Description of Mental Computation
Traditionally, philosophers have emphasized the mind’s representational capacity as

one of its most important features. Perception, motor control, navigation, decision-
making, language acquisition, problem solving, and many other core mental
activities crucially involve representational mental states. For example:
– Perception. Perceptual states represent the environment as being a certain way.
They represent shapes, sizes, colors, locations, and other properties of distal
objects. They are evaluable as accurate or inaccurate, depending on whether
perceived objects actually have the represented distal properties (Burge 2010;
Peacocke 1992).
– Motor control. The motor system transforms intentions into motor commands.
When all goes well, the resulting motor commands promote fulfillment of the
operative intention. For example, if I form an intention to pick up a nearby
ball, then my motor system issues motor commands that hopefully result in my
picking up the ball.
– Navigation. We routinely navigate through the physical environment. In many
cases, we do so by estimating the environment’s spatial layout and by planning a
route to some destination (Evans 1982). Estimates of spatial layout are evaluable
as accurate or inaccurate. Representations of my desired destination are evaluable
as fulfilled or thwarted, depending on whether I reach the destination.
Perception, motor control, and navigation crucially involve mental states with
veridicality-conditions. So do numerous other core mental processes.
Cognitive scientists offer explanatorily successful theories that describe mental
activity in representational terms:
– Perceptual psychology studies how the perceptual system transits from proximal
sensory stimulations (e.g. retinal stimulations) to perceptual states that estimate
shapes, sizes, colors, locations, and other distal properties (Palmer 1999).
Perceptual modeling individuates perceptual states through their representational
properties—as estimates of specific distal shapes, sizes, locations, and so on
(Burge 2010; Rescorla 2015a).
– Sensorimotor psychology studies how the motor system converts intentions into
motor commands that promote fulfillment of those intentions (Rosenbaum 2002;
Shadmehr and Mussa-Ivaldi 2012). The science presupposes that individuals

form intentions with fulfillment-conditions (Jeanerrod 2006; Pacherie 2006;
Rescorla 2016a).
– Beginning with Tolman (1948), many cognitive psychologists have postulated
that mammals navigate using cognitive maps (Gallistel 1990; O’Keefe and
Nadel 1978). Mammals update their cognitive maps based on sensory input and
self-motion cues. Cognitive maps represent spatial aspects of the environment,
including landmark locations as well as the individual’s own current location
(Rescorla 2009, in press).
In these areas, and many others, cognitive science describes how representational
mental states interact with one another, with sensory inputs, and with motor outputs.
A psychological theory that cites representational aspects of mentality is often called
intentional psychology.
Recently, Bayesian cognitive science has elevated intentional psychology to new
heights of mathematical rigor, precision, and explanatory power. The basic idea is
to model mental activity using tools of Bayesian decision theory:
– According to Bayesian perceptual psychology, the perceptual system executes
an unconscious Bayesian inference from proximal sensory stimulations to per-
ceptual states that estimate distal conditions (Feldman 2015; Knill and Richards
1996; Rescorla 2015a).
– According to Bayesian sensorimotor psychology, the sensorimotor system selects
motor commands through unconscious Bayesian inference and decision-making
(Bays and Wolpert 2007; Rescorla 2016a).
– Bayesian models of navigation posit Bayesian updating over cognitive maps that
represent the spatial environment (Cheng et al. 2007; Madl et al. 2014; Madl
2016; Penny et al. 2013; Rescorla 2009).
On a Bayesian approach, the individual (or her subsystems) assigns probabilities to
“hypotheses” drawn from a hypothesis space. Bayesian models typically individuate
hypotheses in representational terms—as representations of specific distal shapes,
sizes, colors, locations, and so on. Bayesian cognitive science describes how
representational mental states (probability assignments to hypotheses that represent
the world) interact with one another, with sensory inputs, and with motor outputs.
The past century has witnessed successive waves of anti-representationalist
sentiment. Advocates of behaviorism (Skinner 1938), connectionism (Ramsey
2007), eliminative materialism (Churchland 1981; Quine 1960; Stich 1983),
interpretivism (Davidson 1980; Dennett 1971), embodied cognition (van Gelder
1992), and dynamical systems theory (Chemero 2009) frequently reject intentional
psychology as unscientific, unconfirmed, unexplanatory, vacuous, or otherwise
problematic. Anti-representationalists recommend that scientific psychology
eschew representational discourse, relying instead upon stimulus-response
psychology, or neuroscience, or some other non-representational scientific
framework. In many cases, anti-representationalists launch highly abstract
philosophical critiques of intentional psychology (Dennett 1971; Quine 1960;
16 M. Rescorla
Stich 1983). I think that anti-representationalism has dramatically failed. Anti-

representational theories have repeatedly shown themselves unequipped to explain
even very basic mental phenomena that intentional psychology readily explains.
Abstract philosophical critiques of intentional psychology tend to be much less
convincing than the representationalist theorizing they purportedly undermine.
I henceforth assume that intentional psychology illuminates perception, motor
control, navigation, decision-making, and many other core mental phenomena. We
reap substantial explanatory benefits by describing these phenomena in representa-
tional terms.
2.4.2 Syntactic Description of Mental Computation
Should syntactic description of mental activity likewise play an important role in

cognitive science?
Fodor (1975, 1987, 1994, 2008) holds that mental computation manipulates items
drawn from the language of thought—an internal system of mental representations.
Mental representations have formal syntactic properties, i.e. properties individuated
without regard to representational import. Mental computation is sensitive to formal
syntactic properties but not representational properties. Fodor holds that a complete
scientific psychology should delineate intentional laws, which describe how mental
states as individuated representationally interact with one another, with sensory
inputs, and with motor outputs. Intentional laws are implemented by computations
describable in syntactic terms. On Fodor’s picture, syntactic manipulation of mental
representations ensures that mental computation transits appropriately between
representational mental states. Fodor also recognizes that any complete cognitive
science will assign a prominent role to neuroscientific description. In this way, he
applies Sect. 1’s three-level picture to mental computation. Representational mental
activity is implemented by syntactic manipulations, which are physically realized
by neurophysiological processes.
Chalmers (2011, 2012) espouses a similar three-level picture of mental activ-
ity, although he places less emphasis than Fodor on representational aspects of
psychological explanation. Field (2001) and Stich (1983) embrace the syntactic
and hardware levels while rejecting the representational level. They hold that
cognitive science should describe mental computation syntactically while ignoring
representational aspects of mentality. Fodor, Chalmers, Field, and Stich all agree
that syntactic description of mental computation should figure crucially within any
complete cognitive science.3
3
Piccinini (2015) assigns a central role to non-representational, multiply realizable descriptions
of artificial and natural computation, including mental computation. He declines to call these
descriptions “syntactic.” Nevertheless, the worries developed below regarding syntactic description
of mental computation also apply to Piccinini’s approach. For further discussion, see (Rescorla
2016b).
Fodor (1981, 2008) maintains that cognitive science already prioritizes syntactic
description of mental activity. I disagree. Contrary to what Fodor suggests, formal
syntactic description does not figure in current scientific theorizing about numer-
ous mental phenomena, including perception, motor control, deductive inference,
decision-making, and so on (Rescorla 2012, 2014b; 2017). Bayesian percep-
tual psychology describes perceptual inference in representational terms rather
than formal syntactic terms (Rescorla 2015a). Bayesian sensorimotor psychology
describes motor control in representational terms rather than formal syntactic
terms (Rescorla 2016a). There may be some areas where cognitive science offers
syntactic explanations. For example, certain computational models of low-level
insect navigation look both non-neural and non-representational (Rescorla 2013b).
But formal syntactic description is entirely absent from many core areas of cognitive
science.
Plausibly, one always can describe mental activity in syntactic terms. The ques-
tion is whether one thereby gains any explanatory benefits. There are innumerable
possible ways of taxonomizing mental states. Most taxonomic schemes offer no
explanatory value. For instance, we can introduce a predicate true of precisely those
individuals who believe that snow is white or who want to drink water. However, it
seems unlikely that this disjunctive predicate will play any significant explanatory
role within a finished cognitive science. Why expect that syntactic taxonomization
will play any more significant a role?
To focus the discussion, consider Chalmers’s functionalist conception of syntax.
Given a true representational or neurophysiological theory of a mental process,
we can abstract away from representational and neural details to specify a causal
topology instantiated by the process. But why suspect that we thereby gain any
explanatory benefits? We can abstract away from a true scientific theory of any
phenomenon to specify a causal topology instantiated by the phenomenon. In most
cases, we do not thereby improve our explanations of the target phenomenon.
Here is a non-psychological example. The Lotka-Volterra equations are first-
order nonlinear differential equations used in ecology to model simple predator-prey
systems (Nowak 2006):
dx
dt
D x .a by/
.LV/
dy
dt
D y .dx c/
where x is prey population level, y is predator population level, t is time, ax is

prey reproduction rate, bxy is the rate at which predators eat prey, cy is predator
death rate, and dxy is predator reproduction rate. Lotka (1910) introduced LV in
order to model oscillating chemical reactions. Researchers have subsequently used
LV to model epidemics (Kermack and McKendrick 1927), economic interaction
(Goodwin 1967), combustion (Semenov 1935), and other unrelated phenomena.
So LV applies not just to ecological systems but also to diverse non-ecological
systems provided that we reinterpret x and y as suitable non-ecological variables.
18 M. Rescorla
These diverse systems instantiate the same causal topology. We can specify their
shared causal topology more explicitly by taking LV’s Ramsey sentence, thereby
suppressing all ecological details.
Ecologists explain predator/prey population levels by using LV where x and
y are interpreted as prey and predator population levels. We do not improve
ecological explanation by noting that LV describes some chemical or economic
system when x and y are reinterpreted as chemical or economic variables, or
by supplementing LV with LV’s Ramsey sentence.4 What matters for ecological
explanation are the ecological interactions described by LV, not the causal topology
obtained by suppressing ecological details. That some ecological system shares
a causal topology with certain chemical or economic systems is an interesting
coincidence, not an explanatory significant fact that illuminates population levels.
The causal topology determined by LV is not itself explanatory. It is just a byproduct
of underlying ecological interactions described by LV when x and y are interpreted
as prey and predator population levels.
Cognitive science describes causal interactions among representational mental
states. By suppressing representational and neural properties, we can specify a
causal topology instantiated by mental computation. But this causal topology looks
like a mere byproduct of causal interactions among representational mental states.
In itself, it does not seem explanatory. Certainly, actual cognitive science practice
does not assign an explanatorily significant role to abstract descriptions of the
causal topology instantiated by perception, motor control, or numerous other mental
phenomena.
Philosophers have offered various arguments why cognitive science requires
syntactic description of mental activity. I will quickly address a few prominent
arguments. I critique these and other arguments more thoroughly in (Rescorla 2017).
Argument from Computational Formalism (Fodor 1981, p. 241; Gallistel
and King 2009, p. 107; Haugeland 1985, p. 106) Standard computational for-
malisms found in computability theory operate at the syntactic level. We can model
the mind as a computational system only if we postulate formal syntactic items
manipulated during mental computation.
Reply The argument misdescribes standard computational formalisms. Contrary to
what the argument maintains, many standard formalisms are couched at an abstract
level that remains neutral regarding the existence of formal syntactic items. We
can construe many standard computational models as defined over states that are
individuated representationally rather than syntactically. Computational modeling
per se does not require syntactic description. My previous writings have expounded
this viewpoint as applied to Turing machines (Rescorla 2017), the lambda calculus
(Rescorla 2012), and register machines (Rescorla 2013a).
4
See (Morrison 2000) for further examples along similar lines.
Argument from Causation (Egan 2003; Haugeland 1985, pp. 39–44) Repre-
sentational properties are causally irrelevant to mental activity. Thus, intentional
psychology cannot furnish causal explanations. We should replace or supplement
intentional psychology with suitable non-representational descriptions, thereby
attaining genuinely causal explanations of mental and behavioral outcomes.
Reply The argument assumes that representational properties are causally irrele-
vant to mental activity. This assumption conflicts with pre-theoretic intuition and
with scientific psychology (Burge 2007, pp. 344–362), which both assign represen-
tational aspects of mentality a crucial causal role in mental activity. We have no
good reason to doubt that representational properties are causally relevant to mental
activity. In (Rescorla 2014a), I argue that indigenous representational properties
of a computing system can be causally relevant to the system’s computations.
Since mental states have indigenous representational properties, it follows that
representational properties can be causally relevant to mental computation.
Argument from Implementation Mechanisms (Chalmers 2012; Fodor 1987,
pp. 18–19) We would like to describe in non-representational terms how the
mind reliably transits between representational mental states. In other words, we
would like to isolate non-intentional implementation mechanisms for intentional
psychology. We should delineate a syntactic theory of mental computation, thereby
specifying non-intentional mechanisms that implement transitions among represen-
tational mental states.
Reply I agree that we should isolate non-intentional implementation mechanisms
for intentional psychology. However, we can take the implementation mechanisms
to be neural rather than syntactic (Rescorla 2017). We can correlate representational
mental states with neural states, and we can describe how transitions among neural
states track transitions among representationally-specified states. As indicated
above, we do not yet know how to do this. We do not yet know the precise
neural mechanisms that implement intentional mental activity. In principle, though,
we should be able to isolate those mechanisms. Indeed, discovering the neural
mechanisms of cognition is widely considered a holy grail for cognitive science.
What value would mental syntax add to an eventual neural theory of implementation
mechanisms?
Argument from Explanatory Generality (Chalmers 2012) Syntactic description
prescinds from both representational and neural properties. Thus, it offers a degree
of generality distinct from intentional psychology and neuroscience. This distinctive
generality provides us with reason to employ syntactic description. In particular,
a syntactic theory of implementation mechanisms offers advantages over a neural
theory of implementation mechanisms by supplying a different degree of generality.
Reply The argument relies on a crucial premise: that generality is always an
explanatory virtue. One can disambiguate this premise in various ways, using
different notions of “generality.” I doubt that any disambiguation of the premise
will prove compelling. As Potochnik (2010, p. 66) notes, “Generality may be of
20 M. Rescorla
explanatory worth, but explanations can be too general or general in the wrong
way.” One can boast generality through disjunctive or gerrymandered descriptions
that add no explanatory value to one’s theorizing (Rescorla 2017; Williamson 2000).
To illustrate, suppose we want to explain why John failed the test. We might note
that
John did not study all semester.
Alternatively, we might note that
John did not study all semester or John was seriously ill.
There is a clear sense in which the second explanation is more general than
first. Nevertheless, it does not seem superior. One might try to disbar such
counterexamples by saying that generality is a virtue when achieved in a non-
disjunctive or non-gerrymandered way.5 But then one would need to show that
syntactic description is itself non-disjunctive and non-gerrymandered, which carries
us back to the question whether syntactic description is explanatorily valuable.
Thus, I doubt that generic methodological appeals to explanatory generality support
syntactic modeling of the mind.6
Overall, philosophical discussion of mental computation has vastly overem-
phasized formal mental syntax. Certain areas of cognitive science may posit
formal syntactic mental items, but there is no clear reason to believe that mental
computation in general is fruitfully described in syntactic terms.
5
Strevens (2008) offers a detailed theory of explanation based on the core idea that good
explanation abstracts away from as many details as possible. However, his finished theory
significantly compromises that core idea, precisely so as to impugn disjunctive explanations.
Strevens seeks to eliminate disjunctive explanations through a causal contiguity condition on good
explanation (2008, pp. 101–109): when we explain some phenomenon through a causal model,
all the model’s realizers should form a “contiguous set” in “causal similarity space.” He says
that we should pursue greater abstraction only to the extent that we preserve cohesion. He says
that overly disjunctive explanantia violate cohesion, because they have non-cohesive realizers.
Strevens’s causal contiguity condition has dramatic consequences for scientific psychology.
Psychological properties are multiply realizable, so psychological explanations are apparently
realized by processes that form a “non-contiguous set” in “causal similarity space.” Hence, as
Strevens admits (pp. 155–165, p. 167), the cohesion requirement prohibits causal models from
citing psychological properties. This prohibition applies just as readily to syntactic description as
to representational description. So Strevens’s treatment does not provide any support for syntactic
explanation of mental activity. He castigates both syntactic explanation and intentional explanation
as non-cohesive.
6
Potochnik (2010) argues that generality is an explanatory virtue only when it advances the
research program to which an explanation contributes. Theoretical context heavily shapes whether
it is explanatorily beneficial to abstract away from certain details. On this conception, one cannot
motivate syntactic description through blanket appeal to the virtues of explanatory generality. One
would instead need to cite specific details of psychological inquiry, arguing that the generality
afforded by syntactic description promotes psychology’s goals. I doubt that any such argument
will prove compelling.
2.4.3 A Case Study: Mammalian Cognitive Maps
To illustrate the themes of this section, let us consider mammalian cognitive maps.
These have veridicality-conditions. For example, a cognitive map that represents
a landmark as present at some physical location is veridical only if the landmark
is indeed present at that location. Detailed, empirically fruitful theories describe
how mammalian cognitive maps interface with sensory inputs, motor commands,
and self-motion cues. The theories describe computations through which mam-
mals form, update, and deploy cognitive maps. In describing the computations,
researchers cite representational properties that contribute to veridicality-conditions
— e.g. they cite the physical location that a cognitive map attributes to a landmark.
Thus, representational description plays a central role within current theories of
mammalian navigation (Rescorla in press).
Neurophysiological description also plays a central role. In comparison with
other areas of cognitive science, we know a fair amount about the neural under-
pinnings of map-based navigation. For example:
– The rat hippocampus contains place cells, each responding selectively to a
specific spatial location.
– The rat entorhinal cortex contains grid cells, each responding selectively to
multiple spatial locations in the available environment. They are called “grid
cells” because the locations where a given cell fires form a periodic grid that
covers the environment.
Neuroscientists have developed mathematical models describing how place cells,
grid cells, and other such cells support mammalian navigation (Evans et al. 2016;
Giacomo et al. 2011). The models aim to illuminate the neurophysiological mecha-
nisms that underlie formation, updating, and deployment of cognitive maps. To be
sure, we are still a long way from completely understanding those mechanisms.
Conspicuously lacking from current scientific research into mammalian naviga-
tion: anything resembling syntactic description. The science describes navigational
computations in representational terms, and it explores the neural mechanisms that
implement those representationally-described computations. It does not describe the
mechanisms in multiply realizable, non-representational terms. It does not abstract
away from neural details of the mechanisms. On the contrary, neural details are
precisely what researchers want to illuminate. Of course, one might propose that
we supplement representational and neurophysiological description of mammalian
navigation with syntactic description. For example, one might articulate a causal
topology that prescinds from representational and neural details. But we have yet
to identify any clear rationale for the proposed supplementation. Certainly, current
scientific practice provides no such rationale. Taking current science as our guide,
syntactic description of mammalian navigation looks like an explanatorily idle
abstraction from genuinely explanatory representational and neural descriptions.
22 M. Rescorla
2.5 Contrast Between Artificial and Natural Computation
I have drawn a sharp distinction between artificial and natural computing systems.
Syntactic description plays a vital role in mediating between representational
description and physical construction of artificial computing systems. In contrast,
many mental computations are usefully described in representational terms rather
than syntactic terms. Why the disparity? Why is syntactic description so much more
important for artificial computation than natural computation?
Sect. 3 emphasized the crucial pragmatic role that syntax plays within computing
practice. By abstracting away from representational properties, syntactic description
offers a workable blueprint for a physical machine. By abstracting away from
physical properties, syntactic description ignores hardware details that are irrelevant
for many purposes. These are practical advantages that immeasurably advance a
practical task: design and construction of physical machines.
Admittedly, we can imagine a computing practice that eschews syntactic descrip-
tion. However, our own reliance on syntactic description secures important advan-
tages over any such hypothetical practice. To illustrate, suppose an agent designs and
builds a machine to execute the Euclidean algorithm. Suppose the agent describes
his machine in representational terms and hardware terms but not syntactic terms.
Now consider a second machine that has very different hardware but instantiates the
same causal topology. Both duplicates satisfy a common abstract causal blueprint.
This commonality is notable even if the agent does not register it. The agent could
have achieved his computing goals by building the second machine rather than
first. In eschewing talk about syntax, the agent foregoes valuable descriptions that
promote his own computing ends. He does not employ syntactic descriptions, but
he should.
Thus, norms of good computing design ensure a key role for syntactic description
of artificial computing systems. Syntactic description enables pragmatically fruitful
suppression of representational and hardware properties.
No such rationale applies to the scientific study of mental computation. Psy-
chology is not a practical enterprise. Cognitive scientists are not trying build a
computing system. Instead, they seek to explain activity in pre-given computing
systems. Constructing an artificial computing system is a very different enterprise
than understanding a pre-given computing system. That formal syntactic description
advances the practical task of designing and constructing artificial computers
does not establish that it advances the explanatory task of understanding a pre-
given computational system. We have seen no reason to think that suppressing
representational and hardware properties of natural computing systems advances
our study of those systems. We have seen no reason to think that formal syntactic
description adds explanatory value to representational and neural description of
mental computation.
Any artificial computing machine was designed by intelligent agents. Good
design practice dictates that those agents sometimes adopt a syntactic viewpoint
Exploring the Variety of Random
Documents with Different Content
NANDI TERRAE Mo
TVS RVINA PROS
TRAVIT SVMPTV PRO
PRIO RESTITVIT [562]
Decio Marco Venanzio Basilio visse ai tempi di Teodorico [563], ed

alcuni cronografi fissano la prefettura di Basilio all’anno 508. I
restauri dell’arena e del podio si praticarono probabilmente dal
Prefetto della Città poco prima dei giuochi venatorî esibiti da Cillica.
Nel Marangoni [564] si legge: «Il sig. cav. Maffei [565] dice essere stato
scritto che mons. Ciampini possedesse un’iscrizione, in cui facevasi
memoria di un risarcimento del Colosseo fatto da . . . . . Teodorico:
ma che avendo egli pregato mons. Bianchini . . . . questa iscrizione
non si è potuta trovare» [566].
L’iscrizione era verosimilmente uno dei soliti sigilli figulini:
✠ REG. D. N. THEODERICO. FELIX. ROMA ovvero

BONO. ROME [567]
PARTE II.
DAL SECOLO VI AL MEDIO EVO.
CAPITOLO PRIMO.
Il Colosseo — Origine di questa voce.
Sòrte le ostilità fra i Goti e l’Impero d’Oriente, Roma andò soggetta

per venti e più anni a gravissimi mali. Non pare perciò probabile che
in quel tempo il popolo romano pensasse ai giuochi ed ai pubblici
divertimenti. Svanì pian piano l’uso degli spettacoli anfiteatrali; e la
grande e venerabile mole dei Flavî rimase inutile e quasi
abbandonata: così principiò a soffrire gli insulti degli uomini e dei
tempi.
Quando nell’anno 663 l’Imperatore Costantino III venne in Roma,
l’Anfiteatro Flavio conservavasi ancora intatto. Costantino depredò i
bronzi dei romani monumenti: «XII dies in civitate Romana
perseverans (Costantinus), omnia quae erant in aere ad ornatum
civitatis deposuit; sed et ecclesiae sanctae Mariae ad Martyres quae
de tigulis aereis erant discoperuit et in regia urbe cum alia diversa
quas deposuerat direxit» [568].
Costantino III fu dunque causa della mancanza degli oggetti di
bronzo che si è verificata in quasi tutti gli scavi praticati nel nostro
Anfiteatro; e probabilmente fu pure ai tempi di quell’Imperatore che
scomparvero i clipei di bronzo e le coperture delle travi esterne del
velario: anzi, con ogni verosimiglianza, fu egli stesso il rapitore della
famosa statua colossale ricordata da Marziale, e che noi scorgiamo
raffigurata sulle medaglie di Gordiano III, posta sopra un basamento,
quasi di contro alla Mèta Sudante. È vero che lì presso, come scrive
Flaminio Vacca [569], fu rinvenuta una testa colossale di bronzo [570],
rappresentante, secondo il parere di alcuni, Commodo; ma quel
rinvenimento non può fare ostacolo alla supposizione accennata,
perchè quella testa, secondo il giudizio degli scultori, e come leggesi
nel Venuti [571], non corrisponde alle misure del Colosso lasciateci
dagli scrittori antichi.
Se la famosa profezia attribuita al Ven. Beda fosse autentica, e se
l’interpretazione che ne dànno alcuni storici antico-moderni fosse
giusta, noi potremmo ritenere che nel secolo VIII l’Anfiteatro Flavio si
conservasse ancora integro. Ma poichè oggi si dubita dell’autenticità
di quel passo [572], ed è ben diversa l’interpretazione che varî storici
contemporanei ci offrono di esso; non potrà sentenziarsi sull’integrità
dell’Anfiteatro nel secolo VIII, fino a che sulla profezia di Beda non si
sparga nuova luce. Sappiamo nondimeno che in quel secolo la
celeberrima mole dei Flavî incominciò ad esser chiamata, anche da
scrittori serî, indifferentemente Amphitheatrum e Colosseum.
Nell’Itinerario di Einsiedeln [573] leggiamo infatti: «Palatin traiani.
Amphitheatrum»; nel Libro Pontificale invece [574] troviamo:
«Colosseum».
Si disputa fra i dotti se l’Anfiteatro Flavio abbia avuto il nome di
Colosseo per la grandiosità della sua mole, ovvero se questa voce
abbia tratto origine dal vicino Colosso di Nerone [575]; o se finalmente
tal denominazione abbia avuto principio dalla posizione topografica
del monumento: vale a dire, se l’etimologia del Colosseo, il quale
«trovavasi nella regione d’Iside e Serapide», provenga dalla
corruzione della voce Collis Isaeum [576].
Riportiamo le ragioni dei singoli scrittori, ed esaminiamo le loro
sentenze.
Il Donati, il Nardini, il Ficoroni, il Venuti e generalmente tutti i
topografi di Roma fino al Nibby, opinarono che «Colosseo» derivi dal
Colosso di Nerone, che sorgeva prossimo all’Anfiteatro Flavio, e che
quel nome sia stato usato per la prima volta nei secoli barbari.
Questi archeologi vedevano tanto limpida detta derivazione, da
crederne inutile un’opportuna dimostrazione. — Che il Colosso di
Nerone fosse celebre nell’antichità, ce l’attesta il ricordo che, con
segni di vera ammirazione, ce ne trasmisero gli storici ed i poeti; e
più ancora lo deduciamo dalla festa annua che ai 6 di Giugno
celebravasi in suo onore e che ci è stata tramandata dal Calendario
Filocaliano colla frase: Colossus coronatus. Questa festa fu
probabilmente istituita in memoria della dedicazione di quel Colosso
al Sole, allorquando Vespasiano, damnatis sceleribus illius principis,
cioè di Nerone, lo coronò con sette raggi colossali [577]. Ma non
sembra credibile che quei dotti abbiano potuto opinare che
l’Anfiteatro Flavio assumesse il nome di Colosseo nei secoli barbari.
Essi infatti dovean sapere (o almeno dubitarne) che a quei tempi il
Colosso non più esisteva. Io mi permetterei piuttosto congetturare
che quegli scrittori pensassero invece che quella voce fosse un’eco
di un modo volgare antico, venuto in uso ai tempi di Adriano, e
precisamente allorquando quel celeberrimo Colosso Neroniano
venne collocato a pochi passi dall’Anfiteatro.
Di questo trasporto, fatto dal suddetto Imperatore, ce ne trasmise la
memoria Sparziano [578]; e la somma difficoltà dell’impresa e la sua
felice attuazione dovettero senza dubbio lasciare nel volgo una
profonda impressione, la quale potè in seguito influir tanto da far
sostituire nel discorso volgare alla parola Amphitheatrum la voce
Colossus. E ciò potè facilmente avvenire cambiando la frase ire ad
amphitheatrum, in: ire ad colossum; cangiamento il quale avrebbe
dato, in questo caso, origine alla frase (ora un po’ strana, ma forse
allora semplicissima): ad Colossum eo; espressione che, per una
naturale eufonia, potè divenire ad Coloss’eo; e poichè
nell’Anfiteatro Flavio si davano continui spettacoli, e v’era quindi
occasione frequente di usare quella frase, pian piano l’Anfiteatro
Flavio divenne addirittura il Colosseo.
Un caso non simile ma uguale è avvenuto ai tempi nostri. Il teatro
principale di Roma, detto di Apollo, sorse presso la Torre di Nona; e
sebbene questa torre non si possa affatto paragonare al famoso
Colosso di Nerone, pur nondimeno essa diè il nome al teatro; e detta
denominazione fu usata da tutti indistintamente, anche dalle persone
di più alto ceto, in modo, che formatasi dalle due parole un’unica
voce, ognuno per dire: vado al teatro di Apollo, diceva: vado a
Tordinona. Non pare adunque impossibile che anche gli antichi
invece di dire: ad amphitheatrum eo, dicessero: ad colossum eo e
poscia, per eufonia, ad Coloss’eo [579].
Se poi si volesse ricercare nell’antichità un’origine più conforme alla
gravità di quei dotti, potremmo opinare che quel vocabolo si
principiasse ad usare subito dopo effettuato il trasporto del Colosso
a pochi passi dell’Anfiteatro; e che, come da Isis nacque Isaeum, da
Adriano Adrianeum, ecc.; così anche dal Colosso sia nata la voce
Colosseum.
In ogni modo, che l’Anfiteatro Flavio sia stato chiamato dal volgo
Colosseo prima del secolo VIII è un fatto certo; e una prova la
troviamo negli stessi documenti del secolo VIII, nei quali la parola
Colosseo è usata come nome proprio dell’Anfiteatro a tutti cognito.
Il Maffei, il Mazzocchi, il Nibby ed altri ritengono che l’Anfiteatro
Flavio non abbia preso il nome di Colosseo per il Colosso, ma per la
sua colossale mole. Ecco le parole del Maffei [580].
«Questa mirabil mole chiamasi in Roma per tradizione
immemorabile il Coliseo; in latino si trova scritto Coliseum o
Colosseum. Il comune consenso dei moderni scrittori ha già fissato
da gran tempo, che così si denominasse l’Anfiteatro dal popolo,
perchè in poca distanza da esso stesse il Colosso di Nerone: ma
alcune considerazioni io proporrò, perchè altri giudichi se così debba
continuarsi a credere. Il Colosso di Nerone [581] alto 120 piedi, opera
di Zenodoro, fu collocato nel vestibolo della sua Casa aurea.
Abbiamo un epigramma di Marziale [582] per cui si trova Tito d’aver
restituita all’uso pubblico, e convertita in benefizio comune quella
grande parte di Roma che Nerone aveva occupata con la sua casa.
Vediamo in esso, come ov’era prima l’atrio, Tito fece strada, in poca
distanza dalla quale era il Colosso, e vediamo come la venerabil
mole dell’Anfiteatro non fu alzata nel sito dell’atrio, o sia del
vestibolo, ma in quello delle peschiere (stagna Neronis erant), che
dovean certamente essere dal vestibolo assai lontane. Presso
all’Anfiteatro, ov’eran prima orti e passeggi, fece Terme chiamate da
Marziale veloci doni (velocia munera); la ragione appar da Suetonio,
che dice furono edificate in fretta (celeriter extrudis).
«Altre osservazioni ancora par che persuadano rimanesse in non
piccola distanza dall’Anfiteatro il Colosso di Nerone. Fu esso poi
mosso dal suo luogo, e fatto trasportare da Adriano: secondo
Sparziano fu allora dedicato al Sole; ma sappiam da Plinio [583],
damnatis sceleribus illius principis, che ciò era fatto fin dai suoi
tempi, in odio alle scelleraggini di Nerone, e però quando il fece
ristorar Vespasiano, di che parla Suetonio. Commodo poi lo tramutò
di nuovo, fattagli levar la testa con riporvi la sua. Ora dice Sparziano
che nel sito ov’era prima il Colosso, fu poi fatto il Tempio della Dea
Roma (De eo loco in quo nunc templum Urbis est), quale non sarà
certamente stato a ridosso dell’Anfiteatro; anzi convien dire ne fosse
assai lontano, s’è il mentovato da Vittore in region diversa (Templum
urbis Romae). L’istesso autore mette pure in region diversa
dall’Anfiteatro un Colosso, distinto tra gli altri, e di consimil
grandezza, che per quello appunto di cui si parla, par si palesi
dall’aver avuto sette raggi intorno al capo, che lo denotavano sacro
al Sole. Non potè adunque denominarsi l’Anfiteatro da statua, che
non gli era prossima, nè attinente per nessun conto».
La prima parte dell’argomentazione del Maffei si basa chiaramente
sopra un falso supposto. Egli infatti crede che il Colosso di Nerone
fosse assai lontano dall’Anfiteatro, mentre ormai nessuno dubita che
il tempio di Venere e Roma, ossia il templum Urbis di Sparziano,
trovavasi immediatamente di fronte al Colosseo; e quindi sappiamo
di certo il posto ove Adriano collocò il Colosso.
Sicchè è cosa positiva l’opposto di quanto opinava l’illustre storico
Veronese; e il Colosso di Nerone, dedicato al Sole, fu sempre vicino
all’Anfiteatro, e dopo il suo traslocamento trovavasi tanto prossimo
ad esso, che se avesse avute aperte le braccia, avrebbe potuto
quasi toccare colla mano i travertini del Colosseo.
Ma prosegue il Maffei: «che se prossimo ancora fosse stato un
colosso a così vasto e dominante edifizio, anzi che dato il nome è
assai più credibile l’avesse preso: e n’abbiam chiaro l’esempio, ove
riferisce Plinio [584]: vocatur Pompeianus a vicinitate theatri, che un
colosso di Giove, grande come una torre, fatto porre nel Campo
Marzio da Claudio, per esser vicino al teatro di Pompeo, acquistò il
nome di Pompeiano».
Il Colosso di Nerone sorse pur troppo vicinissimo all’Anfiteatro,
eppure non prese il nome di Flavius o Flavianus! Nessuno degli
scrittori antichi ce lo ricorda infatti con questo appellativo.
Gli ultimi due argomenti del Maffei sono i seguenti:
«Che se altri mi richiede, donde adunque originata io pensi tal
denominazione, dirò che da null’altro, se non dal comparir questo
edifizio tra tutti gli altri, quel che era tra le statue un colosso, e
dall’uso antico di chiamar così tutto ciò che eccedesse in grandezza.
Vennemi questo pensiero gran tempo fa nel leggere in Suetonio,
come a tempo di Caligola Esio Proculo per l’insigne ampiezza e
bella forma del suo corpo veniva chiamato Colossero o Colosseo;
come forse in quel luogo deve scriversi: ob egregiam corporis
amplitudinem et speciem Colosserus dictus [585]».
Aggiunti altri esempî consimili, così prosegue;
«Mi accertai del tutto scorrendo poi l’Istoria d’Erchemperto Monaco
dell’edizione di Camillo Pellegrini, replicata ora nel tomo secondo
delle Cose Italiche; perchè due volte in essa chiamasi colosso (forse
è da legger Colosseo) l’anfiteatro di Capua, dove non era
certamente il Colosso di Nerone. Appar però manifestamente, come
si dava tal nome agli anfiteatri dal popolo, per la loro maravigliosa
altezza».
Tralascio gli esempî tolti dalla straordinaria grandezza dei corpi
umani, perchè appunto da questi esempî si fa manifesto che il nome
colosso fu sempre proprio delle statue gigantesche, e che da queste
passò a significar coso di grande mole; e vengo all’ultimo
argomento.
Il monaco Erchemperto chiamò colosso e forse Colosseo l’Anfiteatro
di Capua, ove non era certamente il Colosso di Nerone; ma lo
chiamò così quando l’Anfiteatro Flavio già da tempo dicevasi
Colosseo; e se il suddetto monaco chiamò con questo nome
l’Anfiteatro di Capua, dovè così chiamarlo come appunto un
contadino (che io conobbi mentre egli era al servizio di un mio
amico) soleva chiamare Via Appia qualunque antica via lastricata di
poligoni di lava basaltina [586].
Il Mazzocchi non aggiunge agli argomenti del Maffei che l’autorità di
Esichio. È vero che gli etimologisti greci fanno derivare la parola
κολοσσός dallo sforzo che fa la vista per giungere ad una grande
altezza; ma è pur certo che questo vocabolo κολοσσός e dai Greci e
dai Latini fu costantemente usato ad indicare le statue di
straordinaria grandezza.
Il Nibby finalmente dice di non poter ammettere che l’Anfiteatro
Flavio abbia preso il nome dal Colosso di Nerone, perchè nei tempi
barbari questo non più esisteva. L’opinione del Nibby trova una
risposta nella spiegazione già da me enunciata, e che io immaginai
per poterci rendere ragione del come il Donati e gli altri dotti di sopra
citati abbiano potuto ritenere che l’Anfiteatro Flavio prendesse il
nome di Colosseo dal Colosso di Nerone.
Rimane ad esaminare l’opinione del Corvisieri, il quale crede che la
voce Coliseo abbia tratto origine da Collis Isaeum. Ecco le sue
parole: «.... Nel perdere il suo nome una contrada, quello talvolta
non dispariva del tutto ma rimaneva appiccato ad un monumento
vicino; come avvenne dell’Anfiteatro Flavio che prese nome di
Colliseo da una vicina contrada così detta dall’Iseo sulle falde del
colle Esquilino.... È d’avvertirsi che sì l’una che l’altra lezione [587]
conservano chiare le forme del Collis Ysaeum, vocabolo poi
convertito per eufonia in Collisaeum, il quale, come da per sè suona,
non potè mai appartenere in origine all’Anfiteatro Flavio; ma bensì
ad un tempio della Dea Iside, detto dal colle per la sua giacitura, ed
anche per distinguerlo da qualsifosse altro tempio dello stesso titolo.
L’anonimo Einsidlense, che si vuol vissuto tra l’VIII e il IX secolo,
ebbe occasione di nominare nel suo schema topografico di Roma
l’Anfiteatro Flavio, ma lo disse Amphitheatrum e non già
Collosaeum, nè Colisaeum. Ho esaminato inoltre le leggende dei
SS. Martiri, utilissime a rischiarare la topografia di Roma nel medio
evo, come quelle che in buona parte, secondo la sana critica, si
reputano esercitazioni rettoriche della letteratura monastica di quel
tempo; e non ho mai trovato abbiano detto altrimenti che Anfiteatro
quel luogo, il quale, per essere stato destinato alla morte di tanti
campioni del cristianesimo, ebbero spesso il bisogno di nominare. La
terza regione di Roma fu appunto detta di Iside dal tempio di questa
Dea, che come principal monumento vi dovea figurare prima
dell’impero di Tito e di Nerone. La memoria di questo tempio fu
registrata nelle Mirabilia Romae: Coloseum fuit templum Solis, mire
magnitudinis et pulchritudinis, diversis camerulis adaptatum, quod
totum erat cohopertum ereo celo et deaurato, ubi tonitrua, fulgura, et
coruscationes fiebant, et per subtiles fistulas pluvie mittebantur.
Erant preterea ibi signa supercelestia et planete Sol et Luna que
quadrigiis propriis ducebantur. In medio vero Phebus etc. — Ben
s’intende che il Coloseo nell’età delle Mirabilia più non esisteva,
poichè se ne parla come d’un monumento che fu; e quindi la
descrizione che se ne fa così impropria si deve credere basata sulla
volgare tradizione del popolo, il quale, lontano dai tempi dell’idolatria,
potè facilmente esser tratto a credere come indizio del tempio del
Sole qualche avanzo della sua decorazione che accennava ai
misteriosi simboli del culto Isiaco tra’ quali avean pur luogo il Sole, la
Luna ed altri segni celesti. Dobbiamo aver sempre presente che nel
medio evo si giudicò assai grossamente delle nostre antichità. Rari
sono que’ monumenti, anzi rarissimi, che restarono immuni da un
travisamento. Rispetto al Coloseo, poco ci caglia che non si
scrivesse il giusto: ma basti il vederlo indicato ben diverso
dall’Anfiteatro Flavio, com’è altresì questo del Coloseo. Forse fin dai
tempi di Beda era già crollato il Coliseo, secondo mi par di
raccogliere dall’oscurissimo contesto delle riferite parole; nelle quali
con troppa serietà s’è detto racchiudersi una giocosa predizione di
quel pio scrittore.
«Il Beda parla in quel punto della vana presunzione che ha l’uomo di
non errare, della facilità che ne ha, e della vergogna che gliene
deriva se ne venga convinto. A rafforzare la qual sentenza pare si
valesse di quel vaticinio, che, dato come infallibile e come tale
creduto, egli vedeva a’ suoi tempi smentito dal fatto. Il nome di
Coliseo rimase per lungo tempo attribuito alla contrada, e scomparsi
gli avanzi di quel monumento, passò quindi a distinguere
unicamente il vicino Anfiteatro; e fu la colossale figura di questo, per
cui il popolo, ignaro della vera origine del vocabolo, lo ammodò in
Colosseo. A suggellare ciò che ho detto, adduco la gravissima
testimonianza di Benedetto, canonico di S. Pietro (sec. XII), dalla
quale si conosce come a suo tempo fosse ancora distinto l’Anfiteatro
della contrada, che, come ho detto, prese il nome di Colisseo.
Descrivendo egli l’itinerario del Papa nel tornare il lunedì santo dalla
Basilica Vaticana al Laterano, dice che, giunto all’arco trionfale di
Costantino, divertiva a sinistra ante Amphitheatrum et per sanctam
viam juxta Colliseum [588]; e queste parole c’indicano eziandio
chiaramente la postura del Colliseo sulle pendici dell’Esquilino» [589].
L’argomentazione del Corvisieri si riduce a questo: A levante
dell’Anfiteatro v’è una lacinia dell’Esquilino, sulla quale (secondo il
ch. autore) esisteva un tempio Isiaco, creduto nel medio evo del
Sole. Questo tempio dalla sua elevata posizione, per distinguerlo
dagli altri d’Iside che erano in Roma, fu detto Isaeum collis, dal che
collis Isaeum e finalmente Colliseum e Coliseum; termine per lungo
tempo attribuito alla contrada, e che poi, dal popolo ignaro della vera
origine di quel vocabolo, fu applicato all’Anfiteatro Flavio, perchè lo
vedeva un colosso! La poca sodezza di questa argomentazione è
palpabile: con tutto ciò è bene dimostrarla.
Ritenere che su quella parte dell’Oppio la quale guarda l’Anfiteatro
Flavio, sia esistito un tempio Isiaco, è un vero abbaglio. Non v’ha
infatti chi ignori che quel sito fu occupato primieramente dalla Domus
aurea di Nerone, la quale estendevasi dalla somma sacra via fin
oltre le Terme di Traiano, con tutte le sue parti sontuose, non esclusa
la termale e la magnifica piscina detta oggi le Sette Sale: posizione
determinata con chiarezza da Marziale e da Suetonio, e resa certa
dalle escavazioni fatte in quella zona. Poscia sorse su quell’altura la
casa di Tito; ed il rinvenimento del Laocoonte ricordato da Plinio, in
Titi Imperatoris domo [590], ce l’ha dimostrato fino all’evidenza.
Questa casa però non fu che la parte più nobile della Domus aurea,
assegnata da Vespasiano a Tito, ed estendevasi sull’Oppio.
Finalmente sopra una gran parte della domus Titi furono erette le
Terme di Traiano, le quali si conservano ancora in parte, ma che nel
secolo XVI si trovavano in tanto eccellente stato di conservazione,
che Palladio potè lasciarcene i disegni [591]. Sappiamo inoltre che il
tempio d’Iside e Serapide della IIIª regione fu ben lungi da questa
cima dell’Oppio; e sebbene ad alcuni sembrò vederlo sull’estremo
lembo orientale del colle, pur tuttavia la grande maggioranza degli
archeologi lo ritiene sorto nella valle Merulana, presso la chiesa dei
SS. Pietro e Marcellino, dove in ogni tempo vennero in luce copiosi
monumenti Isiaci. Cade così la maggiore della argomentazione del
Corvisieri, e con essa la conseguenza.
Tuttavia, se piacesse considerare per poco alcune prove addotte da
quell’autore a sostegno della sua tesi, si troverebbero vacillanti
assai. Ed invero, che dire del vaticinio così detto di Beda, e del
passo delle Mirabilia riferiti dal Corvisieri al tempio d’Iside? Per ciò
che riguarda il primo, converrebbe immaginarci il tempio d’Iside della
III regione qualcosa di assai più celebre e grandioso del tempio di
Giove Capitolino o del Pantheon, se il profeta, chiunque si fosse,
fece dipendere da quel tempio le sorti di Roma e del mondo!
Relativamente poi al passo delle Mirabilia, fa di mestieri osservare
che questo è preso dalle Mirabilia breviata et interpolata [592] e che
nella prima edizione della Mirabilia [593] e nella Graphia [594] è scritto:
ante Coleseum templum Solis, e non Coleseum fuit templum Solis.
Leggendo adunque, colle prime edizioni, ante Coleseum templum
Solis, si rende chiaro che il templum Solis (che per il Corvisieri
sarebbe lo stesso che Isaeum) non era nè poteva essere il
Coloseum. Se inoltre il passo delle Mirabilia breviata et interpolata
fosse stato riportato per intero, si sarebbe veduto a colpo d’occhio
che le stesse Mirabilia interpolate distinguono il Colosseo dal tempio.
Il passo infatti chiude con queste parole: Ante vero Coliseum fuit
templum in quo fiebant cerimoniae praedicto simulacro (al Colosso
del Sole). Del resto, il rozzo e molto superficialmente erudito scrittore
ci dà senz’altro la descrizione dell’Anfiteatro Flavio attinta dai
classici. In quel coopertum aereo celo et deaurato vi si scorge
l’esametro di Calpurnio: Balteus en gemmis, en illita porticus auro.
Certatim radiant....; nell’ubi tonitrua, fulgura et coruscationes fiebant,
apparisce il passo di Dione: «Il teatro venatorio percosso dal
fulmine.... quasi che l’acqua che vi cadeva da ambo le parti venisse
assorbita dalla forza dei lampi»; in quel per subtilis fistulas pluviae
mittebantur si rileggono le parole di Seneca: Numquid dubitas, quin
sparsio illa, quae ex fundamentis mediae arenae crescens in
summam amphitheatri altitudinem pervenit, cum intentione aquae
fiat.
Ma perchè andar più oltre colle osservazioni, se l’autore si basa su
di un falso supposto?
Ecco come l’Adinolfi giudicò l’opinione del Corvisieri: «Vi è qualche
erudito che vorrebbe distinguere il Coliseo da Anfiteatro, dicendo
che l’Anfiteatro fosse vicino al Colle Iseo, opinione che ha della
sofisticheria» [595].
Di fronte a queste disparate opinioni, il sagace e prudente lettore
sceglierà quella che gli parrà più verosimile.
CAPITOLO SECONDO.
Il Colosseo nel suo abbandono e poscia convertito in fortezza
feudale.
Dalla metà circa del secolo VI al secolo XI il Colosseo, a quanto

pare, rimase abbandonato. Nessuno scrittore di quel corso di secoli
fa menzione di esso; e perciò qui ci è impossibile colmare tant’ampia
lacuna.
Sennonchè questa lacuna non è soltanto propria dell’Anfiteatro
Flavio, ma è comune a tutti i grandiosi monumenti pubblici di Roma;
come, ad esempio, il Circo Massimo, le Terme di Caracalla, quelle di
Diocleziano, ecc. Nè noi possiamo renderci ragione di un tal fatto, se
non opinando col Nibby che questi monumenti «non ostante che più
non servissero allo scopo a cui erano destinati, e per questo lasciati
dallo Stato in abbandono, tuttavia rimanendo di proprietà pubblica
non fosse stato permesso ai potenti privati di quei tempi di occuparli;
trovando così il perchè della mancanza per tre secoli e mezzo di
documenti pubblici e privati relativi a monumenti di questo genere:
sicchè non ci resta che contemplarne lo stato di completo
abbandono in cui si trovarono in questo periodo».
Per quanto riguarda il Colosseo, possiamo ragionevolmente
supporre che fin dalla cessazione dei ludi gladiatorî la custodia
dell’Anfiteatro cominciasse ad essere trascurata, e che sempre più
proseguisse col rarefarsi degli spettacoli venatorî. A questa
trascuranza, d’altronde legittima conseguenza delle calamitose
vicende di quei tempi, e dello spopolarsi della città, attribuì
Teodorico, sul finir del secolo V, la ruina dei monumenti romani,
come egli stesso dice per bocca di Cassiodoro: Facilis est
aedificiorum ruina incolarum subtracta custodia, et cito vetustatis
decoctione resolvitur quod hominum praesentia non tuetur. La reale
ruina però ebbe principio dopo l’ultimo spettacolo dato da Anicio
Massimo. Il Cancellieri [596] scrisse: «Il popolo romano chiese licenza
a Teodorico di ristorare le mura della città colle pietre dei gradini (del
Colosseo) che si trovavano smosse». Questo fatto, il quale trova un
fondamento nei danni arrecati all’Anfiteatro dall’abominando
terremoto di cui parla Venanzio, e nella giusta deduzione che quel
magistrato (per lo scarso numero degli abitanti di Roma a quel
tempo, e per la mancanza di mezzi proporzionati) abbia restaurato
quanto era allora necessario, vale a dire l’arena ed il podio [597];
questo fatto, dico, non può esser avvenuto che nell’ultimo triennio
della vita di quel re, fra il 523 ed il 526, dopo la lettera di sopra
riferita, nella quale Teodorico mostra la sua ripugnanza per i giuochi
sanguinarî ed il desiderio di abolirli. La quale lettera, e specialmente
la sua chiusa, dovè persuadere abbastanza il popolo romano del
volere del re.
Del completo abbandono dell’Anfiteatro a quel tempo, ce ne fa
testimonianza un cimitero cristiano sviluppatosi appunto nei primi
decennî del secolo VI a pochi passi del Colosseo, di fronte
all’ingresso imperatorio che guarda l’Esquilino [598]. Questo cimitero,
da non confondersi coll’altro, più recente, di S. Giacomo, situato a
contatto del Colosseo dalla parte del Laterano, e che ha salvato
dalla distruzione i cinque cippi terminali dell’area esterna
dell’Anfiteatro, venne in luce negli scavi del 1895. Esso si trovava
allo stesso livello dell’Anfiteatro, ed avea le tombe coperte con
tegole improntate di bolli antichi, in nove delle quali si leggevano
marchi dell’età di Teodorico. Una delle tombe, che dall’iscrizione si
potè giudicare del secolo VII circa, si rinvenne all’altezza di due metri
dall’antico piano dell’Anfiteatro, davanti all’ultimo pilastro orientale
del portico, scoperto a piè del colle. Questo cimitero, storico
documento, dopo tredici secoli di esistenza scomparve sotto il
piccone che sistemava l’attuale via, la quale rasenta il Colosseo.
Lasciato l’Anfiteatro a discrezione del tempo, il primo che dovè
risentirne i danni fu senza dubbio il soffitto ligneo del portico
superiore, il quale pian piano dovè corrompersi, lasciando libere a sè
stesse le colonne che lo sostenevano; e queste, nel violento
terremoto che colpì l’Italia nell’aprile dell’anno 801, e recò a Roma
danni gravissimi (tra i quali la ruina della basilica di S. Paolo),
dovettero precipitare giù per la cavea, e sprofondare nell’ipogeo
dell’arena [599]. Dopo questa catastrofe più che mai trovarono
alimento alla vegetazione piante ed arbusti, che, come scrisse
vivacemente il Tournon: plantant leurs racines dans les interstices
des pierres, avaient pris, sur les rampes ruinées, la place des
spectateurs: fu questo senza dubbio il colmo della flora del
Colosseo!
Quelle caverne e quelle boscaglie dovettero dare, con ogni
verosimiglianza, comodo ricetto ad animali d’ogni sorta, non esclusi i
lupi, i quali, come leggesi in una bolla di Paolo II, fin all’anno 1466,
ancor s’aggiravan di notte presso la basilica Vaticana in cerca di
preda. Corpora fidelium quae humabantur in coemeterio dicti campi
(Teutonico) saepe numero reperta fuissent a lupis exhumata.
Finalmente l’Anfiteatro uscì da questo stato di squallido abbandono,
entrando in una nuova fase.
Sul finire del secolo XI l’Anfiteatro Flavio subì le medesime
vicissitudini che subirono gli altri grandiosi edifici di Roma antica. Gli
Orsini occuparono la Mole Adriana — già nel 985 [600], stata
occupata da Crescenzio Nomentano — per molestare Papa
Giovanni XVI; ed il Teatro di Marcello. I Colonnesi presero possesso
del Mausoleo d’Augusto e delle Terme di Costantino sul Quirinale;
ed il Settizonio di Severo e l’Anfiteatro Flavio vennero occupati dai
Frangipani, discendenti della nobile famiglia Anicia, secondo alcuni,
od originarî di Cori e discendenti dai de Imperio, de Imperatore, de
Imperato, Imperii, secondo altri [601].
E qui cade in acconcio rivolgerci una domanda: fu un utile, ovvero fu
un danno per gli antichi monumenti, l’esser passati nelle mani di
nobili famiglie romane? — Se consideriamo i pubblici monumenti
come cosa che dovea rimanere di pubblico dominio (dei quali,
d’altronde, l’autorità legittima in nome e ad utilità del popolo potea
disporre); e se osserviamo la cosa sotto l’aspetto che i monumenti,
caduti nelle mani dei privati, facilmente possono venir deturpati,
modificati, ed anche parzialmente distrutti; non possiamo lodare tali
atti d’impadronimento. Ma se si rifletta che soltanto i monumenti
posseduti dai nobili; che soltanto i materiali e le decorazioni dei
monumenti distrutti, trasferiti nei musei o adoperati in pubblici usi,
nelle chiese, ecc., si sono potuti sottrarre ai colpi del piccone
demolitore, o agli insulti della barbarie, o alla cieca cupidigia di chi
tutto sacrifica al guadagno; se si rifletta, dico, a tutto questo,
dovremo riconoscere che per i monumenti non fu un vero danno, ma
piuttosto un bene l’esser passati in possesso privato delle nobili
famiglie. Che rimarrebbe oggi della tomba di Cecilia Metella, del
teatro di Marcello, del Pantheon, ecc., se nella barbara età di mezzo
non fossero stati ridotti in fortezze o in case feudali, e l’ultimo in
tempio cristiano? La fine di tante statue colonne ed altri marmi, che
ornarono tanti magnifici edifizî, non sarebbe stata in una fornace?.....
Mi si perdoni questa digressione, e torniamo all’argomento.
Noi abbiamo notizia di un Benedetto Frangipane, che nel secolo V,
essendo Patriarca d’Occidente, ebbe la sua dimora in
Trastevere [602], ove possedeva palazzi, case ed il ponte senatorio: e
nella bandiera del rione Trastevere campeggia ancora il leone degli
Anicî. Sulla pianta del Nolli poi, pubblicata nel 1748, la via che
tuttora si chiama ANICIA, viene denominata VIA FRANGIPANE.
I discendenti di questa famiglia emigrarono successivamente in varî
luoghi; e quei che rimasero in Roma ebbero il loro centro principale
sul Palatino, là proprio dove un tempo dimorarono i Papi, e dove nel
secolo IX sorse l’episcopio di Giovanni VIII. Quest’edificio era a poca
distanza dell’Arco di Tito; ed appunto fra l’Arco e l’episcopio i
Frangipani innalzarono una torre, che i cronisti ricordano come il
luogo più sicuro della curia e della cancelleria ecclesiastica: locus
tutissimus curiae. Questa torre, detta perciò Chartularia, fu innalzata
su i resti di un antico edifizio, e trovavasi a sinistra di chi dal
Colosseo s’avanza verso l’Arco di Tito [603].
Oltre alla torre Chartularia, i Frangipani adoperarono a loro fortezze
gli archi di Tito e di Costantino. Ma la fortezza principale dei
Frangipani era presso il Colosseo; anzi era una parte stessa di
questo Anfiteatro, il quale fu posseduto da questa famiglia fin
dall’anno 1130; e possedevano inoltre in quel rione due corpi di
case. Il primo era sulla piazza di S. Giacomo, il secondo trovavasi
presso l’Arco di Tito. Il Papa Innocenzo II [604], a fine di ripararsi dalla
fiera persecuzione dell’antipapa Anacleto II [605], si rifugiò nelle
fortezze dei Frangipani presso il Colosseo. Il card. d’Aragona, nella
vita di quel Pontefice, scrisse: Ad tutas domos Frangipanum de
Laterano descendit, et apud S. Mariam novam et Chartulariam atque
Colossaeum [606]. Tolomeo Lucchese dice: Recollegit in domibus
Frangepaniorum quae in Coliseo erant. F. Tolomeo, vescovo di
Torcello, contemporaneo, nella storia del suo tempo [607] scrive che
nell’anno 1133 Innocenzo II se recollegit in domibus
Frangipanensium, quae erant infra Colisaeum, quia dicta munitio fuit
tota eorum. I Frangipani ebbero presso il Colosseo due case. In
quale di esse il Pontefice Innocenzo II si ricoverò? Qualche moderno
scrittore opina che si ricoverasse in quella del Colosseo, basando la
sua opinione sulle riferite parole di Tolomeo Lucchese, e dalla frase
infra Colisaeum, usata da altri scrittori. L’Adinolfi è di parere che la
parola «infra» possa interpretarsi abbasso od innanzi al Colosseo;
sicchè il loro detto poco varrebbe a sciogliere il nodo della questione.
Le parole del Lucchese sono più chiare, e sembra indicare la casa
che corrispondeva alla piazza di S. Giacomo e che comunicava col
Colosseo. Ciò non ostante, conchiude, non è da stimare per
certissima, non essendo più case di essi addossate al Colosseo, ma
una solamente.
Dalle parole del vescovo di Torcello si deduce che il Colosseo era
stato cangiato in vera fortezza (munitio), difesa da genti armate e
soldati, e che apparteneva alla famiglia dei Frangipani, quia dicta
munitio fuit tota eorum.
La mole resistette agli attacchi della fazione parteggiante per
l’antipapa, il quale, furente ed acceso di collera, andò a
saccheggiare la Basilica Vaticana, il Patriarchio di S. Maria Maggiore
ed altre chiese di Roma, servendosi delle usurpate ricchezze per
corrompere i Romani, onde farsi da questi sostenere.
Innocenzo II passò in Francia, e vi si trattenne fino alla morte dell’ex
ebreo Anacleto II. Al suo ritorno (il quale avvenne nel 1142), dovè
con sommo suo dispiacere, assistere alla cerimonia della
ripristinazione del Senato Romano e della Repubblica, la quale
occupò il Colosseo e tutte le altre torri e fortezze dei Frangipani,
nonchè quelle tenute dagli altri baroni creduti avversi al governo
popolare [608].
«Spenta la persecuzione fatta da Pietro di Pier Leone (antipapa
Anacleto II), si accese nel popolo romano la brama di ridurre nel
proprio dominio Tivoli ed altre città del Lazio. In sulle prime rimasero
vincitori i Tivolesi, ma poi ebbero la vittoria i Romani, sicchè quelli
domandarono mercè al Pontefice, e l’ottennero. Dispiacque la
concessione ai Romani; e, indignatisi contro Innocenzo, posero in
vigore l’antico Senato. La famiglia Frangipani, che avea accolto nelle
sue fortezze il Pontefice, fu tenuta dal popolo come nemica, e la
torre Chartularia ed il Colosseo caddero in sue mani» [609].
Ma la Repubblica e i partiti popolari sono non di rado violente bufere
che duran poco. Quando i popoli s’avveggono dell’inganno e del
lucroso mestiere dei suoi corifei, dànno un passo indietro e tornano
alla calma, tanto loro proficua e necessaria. Pochi anni dopo [610]
Alessandro III, veduta in fiamme la chiesa di S. Maria in Torre, e la
Basilica di S. Pietro nelle mani di Federico I; e, per le tante insidie
tesegli dall’esercito di quest’Imperatore, trovandosi nella dura
necessità di abbandonare il palazzo Lateranense; insieme ai
cardinali ed ai vescovi discese alle sicure case dei Frangipani presso
S. Maria Nuova, la Torre Chartularia ed il Colosseo: e quivi ogni
giorno s’adunavano le Congregazioni, si trattavano cause e si
davano risposte [611].
«In quell’epoca, dice il Gori [612] il Colosseo divenne la fortezza
tutelare della libertà (sic) pontificia»; e dal Panvinio [613]
apprendiamo che in quell’epoca «il Colosseo comunicò il suo nome
ad una regione di Roma della quale i Frangipani erano i capitani, ed i
cui bandonarii precedevano colle insegne il Papa nel dì
dell’incoronazione».
Alessandro III scomunicò Federico I, e, forse nell’Agosto del 1167,
partì da Roma, per maggior sicurezza, nelle due galere o battelli
armati che aveagli mandato sul Tevere il re di Sicilia, Guglielmo [614].
Verso la fine del pontificato d’Innocenzo III (1216), Pietro Annibaldi,
nipote per parte di donna del suddetto papa Innocenzo III [615], volle
edificare una torre nelle vicinanze dell’Anfiteatro, onde poter
attaccare i Frangipani e far loro abbandonare il Colosseo. Le torri
degli Annibaldi erano sulla sostruzione del tempio di Venere e Roma,
e se ne trova una traccia nella pianta di Leonardo Bufalino.
Ma i Frangipani non rimasero inerti, e dalla torre di Naione [616] e
dallo stesso Colosseo procurarono mandare a vuoto il disegno degli
Annibaldi. Questi però non si scoraggirono, ed il desiderio
d’occupare il Colosseo era il loro sogno dorato [617]; ed ecco che si
presenta loro un’occasione propizia. Federico II si porta in
Acquapendente: si manifesta persecutore della Chiesa; rompe le
relazioni con papa Gregorio IX, e mette in iscompiglio la città di
Roma. L’Imperatore ebbe per un momento il sopravvento; e gli
Annibaldeschi approfittarono di questa congiuntura per ottenere che
Federico II forzasse i Frangipani, Enrico e Giacomo, a ceder loro la
metà del Colosseo coll’annesso palazzo, e a sanzionare la cessione
con giuramento [618]. Forse sull’altra metà aveva diritto il Senato
Romano fin dai tempi di Corrado, allorquando fu violentemente
presa; e ciò per porre nelle mani dei suoi favoreggiatori metà
dell’ampio edificio.
E per giungere a tale determinazione, debbon esser sopraggiunti dei
fatti che noi ignoriamo; poichè Federico II, all’epoca di Gregorio IX,
quando era in possesso di quella fortezza, fu da Pietro Frangipane
molto ben trattato.
I Frangipani, alla lor volta, reclamarono presso Innocenzo IV,
domandandogli l’annullamento di quel trattato. Il papa annuì, e con
breve del 18 marzo 1244 dichiarò nulla la cessione del Colosseo,
per non essere stata opportunamente chiesta dai Frangipani
l’indispensabile facoltà di poter cedere un luogo del quale essi non
eran padroni, ma semplici feudatarî del sovrano Pontefice; e dichiarò
pur nulla la permuta degli altri beni, perchè fatta non con libertà, ma
sotto la violenza e le minacce di Federico II. Ecco il tenore della
bolla: «Quum sicut lecta coram nobis vestra petitio continebat, nuper
apud Aquapendentem in presentia Principis constituti, eidem ad
suam instantiam ipsius timore perterriti, medietatem Colisei cum
palatio exteriore sibi adiacente et omnibus iuribus ad ipsam
medietatem pertinentibus dilecto filio Anibaldo civi romano titulo
pignoris obligata, quae ab Ecclesia Romana tenetis in feudum de
facto cum de iure nequiveretis, duxeritis concedenda, praestitis
nihilominus iuramentis vos contra concessionem huiusmodi non
venturos, licet ex hoc essetis non immerito puniendi, attendentes
tamen, quod coacti quodammodo terrore tanti principis id fecistis,
concessionem huiusmodi nullam esse penitus nuntiantes praedicta
ad vestrum et Ecclesiae Romanae ius et proprietatem auctoritate
praedicta revocamus: iuramentis praedictis nihilominus relaxatis,
eadem auctoritate excomunicationis vinculo, ac poenae quinque
millium marcharum argenti omnes qui contravenire praesumserit
supponentes» [619].
Il Sommo Pontefice (per impedire che il Colosseo andasse a cadere
nelle mani di Federico II, con grave danno di Roma) dichiarò
formalmente esser l’Anfiteatro di diretto dominio della Santa Sede; e
per questa pontificia dichiarazione si vennero a far manifeste le
differenti opinioni dei varî partiti; poichè alcuni credevano che il
Colosseo appartenesse alla Chiesa, mentre altri ritenevano
appartenesse all’Imperatore.
Annullato il contratto, gli Annibaldeschi dovettero abbandonare il
Colosseo, ove in quel frattempo avevano abitato: Annibaldenses
quoque Romani Proceres se munierunt, in Colossaeo, in eoque
habitarunt, quemadmodum antea Frangipanes [620]; e i Frangipani
tornarono nel loro primitivo possesso.
«E quanto alle abitazioni fatte dai Frangipani entro il Colosseo, si
riconoscono fino al presente le muraglie che occupano e dividono fra
gli archi esteriori e gli interiori sopra l’antiche scalinate, al numero di
13 verso il Laterano, onde il circuito era molto considerevole, ed è a
Welcome to our website – the ideal destination for book lovers and
knowledge seekers. With a mission to inspire endlessly, we offer a
vast collection of books, ranging from classic literary works to
specialized publications, self-development books, and children's
literature. Each book is a new journey of discovery, expanding
knowledge and enriching the soul of the reade
Our website is not just a platform for buying books, but a bridge
connecting readers to the timeless values of culture and wisdom. With
an elegant, user-friendly interface and an intelligent search system,
we are committed to providing a quick and convenient shopping
experience. Additionally, our special promotions and home delivery
services ensure that you save time and fully enjoy the joy of reading.
Let us accompany you on the journey of exploring knowledge and

personal growth!
textbookfull.com

PDF Philosophy and Computing Essays in Epistemology Philosophy of Mind Logic and Ethics 1st Edition Thomas M. Powers (Eds.) download

Uploaded by

Copyright:

Available Formats

PDF Philosophy and Computing Essays in Epistemology Philosophy of Mind Logic and Ethics 1st Edition Thomas M. Powers (Eds.) download

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

PDF Philosophy and Computing Essays in Epistemology Philosophy of Mind Logic and Ethics 1st Edition Thomas M. Powers (Eds.) download

Uploaded by

Copyright:

Available Formats

Download the Full Version of textbook for Fast Typing at textbookfull.

Philosophy and Computing Essays in Epistemology

Download More textbook Instantly Today - Get Yours Now at textbookfull.com

Philosophy in Educational Research Epistemology Ethics

Beyond Faith and Rationality Essays on Logic Religion and

The Philosophy of Piers Plowman: The Ethics and

Philosophy of Mind in the Twentieth and Twenty First

Ethics Beyond the Limits New Essays on Bernard Williams

Grammar, Philosophy, and Logic 1st Edition Bruce Silver

Implicit bias and philosophy. Volume 1, Metaphysics and

Logic and Philosophy of Time Themes from Prior 1st Edition

Thomas M. Powers Editor

Executive Editorial Board

Board of Consulting Editors

Philosophy and Computing

ISSN 0921-8599 ISSN 2542-8349 (electronic)

Library of Congress Control Number: 2017952948

© Springer International Publishing AG 2017

Printed on acid-free paper

This Springer imprint is published by Springer Nature

1 Introduction: Intersecting Traditions in the Philosophy

Keywords Philosophy of computing • Cognition • Representation • Autonomy

T.M. Powers ()

© Springer International Publishing AG 2017 1

Keywords Levels of explanation • Representation • Syntax • The computational

2.1 Representation, Syntax, and Hardware

© Springer International Publishing AG 2017 5

2.2 Representational Description of Artificial Computation

Hardware description figures indispensably within computing practice. Ultimately,

2.2.1 Representation Elucidated

2.2.2 The Value of Representational Description

To what extent do we illuminate an artificial computing system by citing its

we are often primarily concerned with representational aspects of computation. The

2.3 Syntactic Description of Artificial Computation

There has been considerable philosophical discussion surrounding the notion of

Even when computer scientists are most fundamentally interested in

representational properties. However, we cannot ensure that other users of some

implement representationally-specified computations. It plays a crucial role in

2.4 Natural Computing Systems

2.4.1 Representational Description of Mental Computation

Traditionally, philosophers have emphasized the mind’s representational capacity as

Shadmehr and Mussa-Ivaldi 2012). The science presupposes that individuals

Stich 1983). I think that anti-representationalism has dramatically failed. Anti-

2.4.2 Syntactic Description of Mental Computation

Should syntactic description of mental activity likewise play an important role in

where x is prey population level, y is predator population level, t is time, ax is

2.4.3 A Case Study: Mammalian Cognitive Maps

2.5 Contrast Between Artificial and Natural Computation

Decio Marco Venanzio Basilio visse ai tempi di Teodorico [563], ed

✠ REG. D. N. THEODERICO. FELIX. ROMA ovvero

Sòrte le ostilità fra i Goti e l’Impero d’Oriente, Roma andò soggetta

Dalla metà circa del secolo VI al secolo XI il Colosseo, a quanto

Let us accompany you on the journey of exploring knowledge and

You might also like