Reason and Less: Pursuing Food, Sex, and Politics

Reason and Less

Pursuing Food, Sex, and Politics

Vinod Goel

The MIT Press

Cambridge, Massachusetts

London, England

© 2022 Massachusetts Institute of Technology

All rights reserved. No part of this book may be reproduced in any form by any electronic or mechanical means (including photocopying, recording, or information storage and retrieval) without permission in writing from the publisher.

The MIT Press would like to thank the anonymous peer reviewers who provided comments on drafts of this book. The generous work of academic experts is essential for establishing the authority and quality of our publications. We acknowledge with gratitude the contributions of these otherwise uncredited readers.

Library of Congress Cataloging-in-Publication Data

Names: Goel, Vinod, author.

Title: Reason and less : pursuing food, sex, and politics / Vinod Goel.

Description: Cambridge, Massachusetts : The MIT Press, [2022] | Includes bibliographical references and index.

Identifiers: LCCN 2021017752 | ISBN 9780262045476 (paperback)

Subjects: LCSH: Decision making. | Reasoning. | Logic. | Cognitive neuroscience.

Classification: LCC BF448 .G64 2022 | DDC 153.4/3—dc23

LC record available at https://lccn.loc.gov/2021017752

d_r0

To the memory of my mother and father

For Kalpna, Amit, and Natasha, who taught me there is often less to life than reason

Contents

Preface

Acknowledgments

I The Rational Animal

1 Food, Sex, Politics, and the Rational Animal

Examples of Reasoning in the Real World

What Is Rationality?

From Rationality to Decision Theory

Rationality in the Real World: Global Warming Example

Rationality in the Real World: Other Examples

Organization of the Book

2 The Enigma of Rationality: Fallen Angel or Risen Ape?

Rationality as God’s Grace

Rationality and Darwin: Nothing Special

Rationality and Cognitive Science: Maybe a Little Bit Special

Rationality: Special but Tethered

II Kinds of Minds

3 Reflexes, Homeostasis, and the Autonomic Mind

4 The Instinctive Mind

What Are Instincts?

Mechanistic Models of Instincts

Are Instincts Enough?

Extended Example: Are Gender-Specific Behaviors Instinctive or Social Constructs?

5 The Associative Mind: More than Instincts, Less than Reason

Classical Conditioning

Behaviorism and Operant Conditioning

Associationism without Behaviorism

Are Associations Enough?

Appendix: Mechanistic Accounts of Associations

6 The Reasoning Mind: Propositional Attitudes and Coherence

Animal Symbolicum

Propositional Attitudes and the Structure of Representational Mental States

Propositions and Coherence

Can My Goldfish Reason?

From the Cognitive Mind to the Computational Mind

Summarizing the Reasoning Mind along the Five Dimensions

Is the Reasoning Mind Enough (or Too Much)?

Appendix: Formal Systems and Information Processing in Physical Symbol Systems

III Reasoning with the Cognitive Mind

7 Logical Inference: Heuristic and Analytical Systems

Cognitive Theories of Logical Reasoning

Content Effects in Reasoning

Dual Mechanism Theories of Reasoning

8 Conceptual Inference in the Real World: From Science to Politics

Conceptual Coherence from Hume to Goodman

Cognitive Approaches to Inductive Reasoning

Inductive Reasoning in the Real World

The Reasoning Mind Recruits the Instinctive Mind

IV The Tethered Mind

9 The Instinctive Mind Resurrected: Modularity, Reciprocity, and Blended Response

Massive Modularity: Instincts All the Way Up

Reciprocity and Cheater Detection in Nonhuman Animals

Reciprocity in Humans: Self-Maximization, Fairness, Cheating, and Punishment

Tethered Rationality: Blend of Instincts and Reason in Cooperative Economic Decision-Making

Appendix: A Conceptual Critique of Massive Modularity

10 Kinds of Brains: Of Mice, Monkeys, and Men

Overview of Brain Evolution

Brain Stem, Diencephalon, and Subcortical Systems: Essential and Mostly Hardwired

Cerebral Cortex: Mostly Softwired

Accounting for the Reasoning Brain

A Tethered Brain for a Tethered Mind

11 Feelings: Chocolate, Lust, and Coherence

Who’s Afraid of Feelings?

Characterizing Feelings

From Feelings to Emotions

Origins of Feelings

Function of Feelings: Motivate and Guide

Feelings and the Autonomic Mind

Feelings and the Instinctive Mind

Feelings and the Associative Mind

Feelings and the Reasoning Mind

12 Control Structures: Who Is in Charge of the Tethered Mind?

Western-Christian Model Endorsed by St. Paul (and God)

Standard Cognitive / Social Science Reasoning Model

Massive Modularity Model

Dual Mechanism Models

Control Structure for Tethered Rationality

Evaluating the Tethered Rationality Model

V What Color Is Your Bubble? Why Changing Minds Is Hard

13 When Failures of Belief Revision Are Less than Motivated Reasoning or Sloppy Reasoning

Explaining Failure of Belief Revision as Motivated Reasoning or Sloppy Reasoning

Motivated Reasoning as Question Begging

Evidence for In-Group/Out-Group Bias Instinct

Tethered Rationality and Failure of Belief Revision

Are False Beliefs and Bubbles Sustainable: Do Facts Matter?

14 Global Belief Revision Is Constrained by Neural Maturation

Local Belief Revision versus Worldview Revision

Brain Development: Overview

Prenatal (Experience-Expectant) Brain Development

Postnatal (Experience-Dependent) Brain Development and Behavior

Consequences for Models of Rationality

VI What Follows from the Tethered Mind?

15 Concerns, Consequences, and Conclusions

Concerns and Consequences

Concluding Notes to My Colleagues

Concluding Notes to the General Reader

Bibliography

Index

List of Figures

Figure 1.1 An example of the rational mind at work using a hypothetical reconstruction of the US decision to invade Iraq in 2003. Each subgoal follows coherently from the preceding goal or subgoal plus beliefs, eventually resulting in an action. The integration of goals and subgoals plus beliefs via the coherence relation is the nexus of the reasoning step.

Figure 1.2 Simple decision tree and utility function. One might model the decision to invade or make a deal as follows. The chances of a successful invasion are 0.75, while the chances of failure are 0.25. The chances of a successful negotiation are 0.15, while the chances of failure are 0.85. The value assigned to both the successful invasion and successful negotiation is +100. The value assigned to a failed invasion and failed negotiation is −100. Based on these values, the utility of invasion is +0.5 and the utility of negotiation is −0.7 (utility = Σ(probabilityoutcome × valueoutcome)). Notice that decision theory provides no guidelines for assigning probabilities of outcomes and the value of the outcomes, but once these numbers are (magically) assigned, simple probability theory allows us to coherently calculate expected utility. The rational choice is the one with the highest expected utility.

Figure 1.3 The carbon bathtub analogy. If more water is dripping into the bathtub than is leaving via the drain, no matter how small the difference, the coherent conclusion is that the bathtub will eventually overflow. This is an example of a basic, intuitive coherency judgment. If one fails to acknowledge it (in good faith), it is not clear what more can be said to change one’s mind. This would constitute a cognitive failure in detecting coherency.

Figure 3.1 A monosynaptic reflex arc. The sensory input (tapping) generates the motor response directly through a monosynaptic reflex arc. There are no intervening neurons between the sensory nerves communicating the tapping signal from the quadricep muscles back to the spinal cord, and the motor neurons from the spinal cord innervating quadricep muscles. The figure also shows the presence of an inhibitory interneuron that serves to relax the hamstring muscle. Drawing by Aldona Griskeviciene.

Figure 3.2 Control of respiration by the autonomic system, in part by a series of reflex actions with control centers in brain stem structures.

Figure 3.3 Homeostatic control of blood glucose levels by the autonomic system.

Figure 4.1 The Lorenz hydraulic or energy model of instincts. Adapted from Lorenz (1950).

Figure 4.2 Differentiation of sex and gender in mammals.

Figure 5.1 (a) All animals can make contiguity-based associations. (b) Only humans can make similarity-based associations. (c) Similarity-based associations can lead to reasoning. Based on drawings by William James (1878).

Figure 5.2 Fragment of a semantic network representation of dog.

Figure 5.3 (a) The biology of neural networks. (b) Axiomatization of neural networks: converting a biological problem into a computational problem. (c) Training of neural networks. Modeled after Rumelhart and McClelland (1986).

Figure 6.1 A transitive reasoning schema modeled after Halford, Wilson, & Phillips (2010).

Figure 9.1 Wason card selection task.

Figure 10.1 Example of a phylogenetic tree. Some animal illustrations are reproduced from Angel Cabrera (1919).

Figure 10.2 Simplified phylogenetic tree of brain development across reptiles, birds, and mammals. There are five things to note: (1) the brain stem, cerebellum, and thalamic structures appear early and are conserved across the evolutionary tree; (2) the telencephalon begins to differentiate into the pallium and subpallium with reptiles; (3) the subpallium evolves into the basal ganglia structure of mammalian brains; (4) the pallium starts as a largely undifferentiated structure in reptiles and becomes more differentiated in birds and even more fully differentiated into subcortical and cortical structures in mammals; and (5) in mammals, the pallium becomes the six-layered neocortex.

ADVR = anterior dorsal ventricular ridge; Ac = accumbens; B = basorostralis; Cd = caudate; E = entopallium; GP = globus pallidus, internal (i) and external (e) segments; HA = hyperpallium apicale; Hp = hippocampus; IHA = interstitial hyperpallium apicale; L2 = field L2; MD = dorsal mesopallium; MV = ventral mesopallium; OB = olfactory bulb; PDVR = posterior dorsal ventricular ridge; Pt = putamen. For birds, there is some uncertainty as to whether the MD is hyperpallium densocellulare (HD) or a separate structure. Reproduced with permission (with slight modifications) from Jarvis (2009).

Figure 10.3 The major components of the human brain starting with (a) brain stem and cerebellum, (b) diencephalon (thalamus and hypothalamus), (c) subcortical structures, including basal ganglia nuclei, hippocampus, and amygdala, (d) and cerebral cortex. Figure drawn by Brooklyn McKinley.

Figure 10.4 Brain systems are hierarchically organized and tethered. This example from the motor system highlights interconnections and coordination across four levels of brain structures (brain stem and cerebellum, thalamus, subcortical structures, and cerebral cortex). Most brain functions require such system integration and coordination. Notice that information flows both ways, top-down and bottom-up. Figure modeled after information in Sherwood and Kell (2009).

Figure 10.5 Brain stem, diencephalon, and subcortical systems involved in aggression, dominance, and social attachment behaviors in rodents. AMY = amygdala; AOB = accessory olfactory bulb; BNST = bed nucleus of the stria terminalis; DRN = dorsal raphe nucleus; HPC = hippocampus; HyP = hypothalamus; IL = infralimbic division of the mPFC; LHb = lateral habenula; LS = lateral septum; MeA = medial amygdala; MOB = main olfactory bulb; MOE = main olfactory epithelium; mPFC = medial prefrontal cortex; NAc = nucleus accumbens; OFC = orbitofrontal cortex; PAG = periaqueductal gray; PL = prelimbic division of the mPFC; VMHv1 = ventrolateral subdivision of the ventromedial hypothalamus; VNO = vomeronasal organ; VTA = ventral tegmental area. Reproduced with permission from Ko (2017).

Figure 10.6 Developmental timeline of various brain structures in rhesus monkeys is consistent with the old adage ontogeny recapitulates phylogeny. Neurogenesis and maturation occur from the inside out, beginning with the brain stem (monoamine cells), diencephalon, and subcortical structures and then cortical structures. E indicates embryonic days. Gestation period is 165 days. Data compiled and graphed by Selemon and Zecevic (2015) from studies by Rakic and colleagues. Reproduced with permission.

Figure 10.7 Brain systems and neural network properties associated with the four behaviors of interest. The qualitative differences in the four behaviors or minds under consideration are underwritten by the appearance and variability of brain structures in the phylogenetic tree and their corresponding levels of experience-expectant and experience-dependent maturation schedules. Innate structural constraints apply to certain parts of the cortex, so they exhibit the properties of compositionality, systematicity, productivity, and inferential coherence required by the reasoning mind. The tethering of the various behaviors and the underlying neural systems allows for a single blended or integrated response.

Figure 11.1 (a) The location of the hedonic hotspot discovered by Olds and Milner (1954) in the septal nucleus of rats has been reconstructed and found to be very close to the nucleus accumbens. (b) Placement of electrodes in or near the nucleus accumbens in one patient reported by Heath (1972) resulted in feelings of sexual arousal. (c), (d) Hedonic reward systems in both rodents and humans involve similar interlinked brain stem, diencephalon, and subcortical networks, with some cortical representation, particularly in humans. The systems involved include the periaqueductal gray (PAG), the ventral tegmental area (VTA), ventral pallidum, nucleus accumbens, amygdala, hypothalamus, insular cortex, cingulate cortex, and orbital frontal cortex. (e) Shows the identification of distinct liking/disliking and wanting hotspots in the nucleus accumbens of a rat brain. (f) Reward regions are represented in the human orbital frontal cortex. Figure reproduced (with some reorganization) from Kringelbach and Berridge (2010) with permission of the authors.

Figure 11.2 The rat’s system for taste discrimination. Chemical reactions in the oral cavity between food and the taste buds are recognized as distinct tastes in the insular cortex after processing in the brain stem and thalamus. GG = geniculate ganglia; IC = insular cortex; NG = nodose ganglia; NST = nucleus of the solitary tract; PbN = parabrachial nucleus; PG = petrosal ganglia; VPMpc = ventral posterior medial nucleus of the thalamus. Figure reproduced with permission from Matsumoto (2013).

Figure 11.3 The top graph tracks the pleasure modulation associated with the appetitive (wanting), consummatory (liking), and satiety (equilibrium) phases of the LUST system in humans. The brain images show the activation of brain regions as a function of three phases of pleasure. Both subcortical and cortical regions are involved. Given the ubiquity of the LUST system in large segments of the phylogenetic tree, the generators of the feelings and behaviors will be in the brain stem, diencephalon, and subcortical regions, while their representations also involve cortical regions. Reproduced with permission from Georgiadis and Kringelbach (2012).

Figure 12.1 Various models of control structure found in the literature. (a) This is the control structure endorsed by St. Paul and God and is still the basis of much Western societal, religious, and legal norms. (b) This is the generic control structure of the standard cognitive and social science model. Reason is the only route to action. (c) The massive modularity model has no reasoning component. The action is determined by a set of interacting instincts. (d) The flow of control for the default interventionist model advocated by many adherents of dual mechanism theories. (e) The parallel competitive model advocated by other adherents of dual mechanism theory accounts.

Figure 12.2 Control structure for tethered rationality. Human behavior is a function of the responses of the autonomic system, the instinctive system, the associative system, and the reasoning system to any given situation. While the mechanism underlying each system is different, all utilize the common currency of feelings. The response generated by each system is also in the currency of feelings, with valence, arousal, and duration components. The system is set up to maximize pleasure and minimize pain or displeasure. The selected behavior will usually be a blended response based on the output of all systems. There is no central executive in charge. The reasoning system has an input into the response, but so do the other systems. Individual differences in behavior are explained not just in terms of individual differences at the level of reasoning but also individual differences at the level of the autonomic, instinctive, and associative systems. A notion of variable individual effort may be captured in this model by variability in levels of arousal associated with different systems.

Figure 12.3 Tethered rationality control structure for human energy management. Energy management is reasonably well understood, though details continue to be refined and updated. It is in part a homeostatic system sensitive to nutrient signals, gastrointestinal signals, adiposity signals, and signals from other organs. The dashed lines indicate homeostatic processes. CCK = cholecystokinin; GI = gastrointestinal; GLP-1 = glucagon-like peptide-1; NPY/AgRP = neurons expressing neuropeptide Y/ agouti-related protein; PAG = periaqueductal gray; POMC = neurons expressing pro-opiomelanocortin; PYY3−36 = peptide YY3−36; VTA = ventral tegmental area.

Figure 13.1 Process of belief revision under different models. As new evidence comes in, prior beliefs need to be updated and revised via the coherence relation. The models vary in terms of goal directedness or vested interest of the reasoner and how these interests and prior beliefs impact the filter. The purpose of the filter is to select and weigh the evidence. (a) This is reasoning in an ideal world. The reasoner is disinterested in the outcome, the filter lets through all relevant information, and the coherence relation revises beliefs accordingly. While no individual human can implement this model, scientific methodology over time converges toward it. (b) This model represents best practice real-world reasoning. The reasoner tries to maintain a disinterest in the outcome and minimize the impact of prior beliefs on the filter. This is indicated by the dashed line connecting prior beliefs to the filter. (c) The motivated reasoning model is characterized by goal directedness and a robust, active impact of prior beliefs on the filtering of evidence in the service of a directed goal. This can result in the shielding of belief systems from evidence. (d) In the sloppy reasoning model, the emphasis is on some sort of disruption in the determination of the coherence relation itself, indicated by the dashed circle, be it through lack of effort, intelligence, or succumbing to common reasoning fallacies, such as the confirmation bias, a natural tendency to look for confirming evidence and ignore conflicting evidence.

Figure 14.1 Important stages in neural development and maturation. Graphed from data from Andersen (2003) and Stiles and Jernigan (2010).

Figure 14.2 Time course of synaptogenesis and synaptic pruning for sensorimotor, language, and higher cognitive systems. The general time course of gliogenesis and myelination is also indicated. Figure based on Casey, Tottenham, Liston, & Durston (2005), modified based on data from Gogtay et al. (2004).

List of Tables

Table 6.1 Kinds of minds. A summary of how the four behaviors of interest—autonomic, instinctive, associative, and rational—are distinguished along five dimensions. The inclusion of earlier evolved brain structures in more recently evolved behaviors is meant to foreshadow the tethering explicitly discussed in chapter 10.

Table 11.1 Panksepp’s primordial emotions (or instincts)

List of Boxes

Box 6.1 Illustration of recursion

Box 10.1 Phylogenetic tree

Preface

After more than 20 years of studying the neural basis of rationality, it dawned on me that there was very little consequential human behavior that I could explain. Nothing I have learned about rationality was relevant to understanding my teenage daughter. Nothing I have learned about rationality is relevant to explaining the behavior of my MAGA (Make America Great Again) Florida friends and neighbors who profess an unshakable faith in American exceptionalism (which I accept and have benefited from) but then deny and ridicule the sciences of vaccines and climate change emerging from exceptional American institutions. Nothing I have learned about rationality seems particularly relevant to explaining certain views of my ultraliberal friends and colleagues, such as gender being just a social construct, despite scientific evidence to the contrary. Nothing I have learned about rationality is relevant to explaining why intelligent, powerful men engage in sexual indiscretion, even assault, at great personal risk and harm to others. Nothing I have learned seems particularly relevant to explaining why I overindulge in chocolate cake and pizza, despite being overweight. Based on the standard models of reasoning, the only explanatory tools available are appeals to heuristics, some form of motivated reasoning, poor education, or perhaps cognitive deficiency. Such explanations may apply in specific individual cases, but they cannot account for all or even much of human behavior. I have come to believe that we are making a fundamental mistake in bringing only the tools of rationality to explain human behavior.

My main message is that, while we are rational animals, explaining real-world human behavior just in terms of reasoning does not get us very far. We have to recognize that nonreasoning systems also affect actual behavior. We need to look beyond (or below) reason to noncognitive factors to fully account for human behavior. Much human behavior that does not conform to our expectations of rationality is not irrational but rather arational, by which I mean that it is not reason based. Some nonreasoning systems are initiating and/or modulating the behavior.

The goal of this book is to undertake a commonsense reconsideration and recalibration of theories of human behavior. Human behavior needs to be explained in terms of the workings of autonomic systems, instinctive systems, associative systems, and reasoning systems. Each of these systems has been extensively studied. How these systems communicate and interact to account for human behavior is rarely considered. I sketch out a proposal that I call tethered rationality, in which human behavior is a blended response incorporating inputs from each of these systems. The challenges are to provide empirical data for the blended response hypothesis, show how the tethering is supported by the neurophysiology, propose a common currency that would allow these systems to communicate and interact, and provide a control structure for the overall system. Meeting these challenges takes us on a fascinating journey through psychology (cognitive, behavioral, developmental, and evolutionary), neuroscience, philosophy, ethology, economics, and political science, among other disciplines.

One key insight that holds the model together is that feelings—generated in old, widely conserved brain stem structures—are evolution’s solution to initiating and selecting all behaviors and provide the common currency for the four different systems to interact. Reason is as much about feelings as is lust and the taste of chocolate cake. All systems contribute to behavior and the overall control structure is one that maximizes pleasure and minimizes displeasure. Such an account drives human behavior back into the biology, where it belongs, and provides a richer set of tools to understand how we pursue food, sex, and politics.

Models not only explain behavior but also have consequences for changing it. The model of tethered rationality is no exception. For those engaged in changing behaviors—such as sexism, racism, cheating, or even climate change denial—tethered rationality may have the unwelcome message that such behaviors cannot be easily changed by changing beliefs through a few days of sensitivity training. This is not to say that they cannot be changed at all, but rather that more drastic measures will be required, the nature of which will depend on the specific behavior in question. Having an accurate model of human behavior is the first step in this endeavor.

Utopia, Ontario, Canada

May 2021

Acknowledgments

I would like to thank the following colleagues from diverse areas of expertise for reading one or more chapters of the manuscript and offering valuable feedback: Ellen Bialystok, Christopher von Bülow, Ron Chrisley, Hugo Critchley, Wim De Neys, Shira Elqayam, Larry Fiddick, Jordan Grafman, Sam Gilbert, Ira Novak, Magda Osman, Jerome Prado, Steven Sloman, and Hongkui Zeng. Among friends and students, I’m grateful to Mark Hewitt, Ron Bean, Adam Burnett, and Claire Quenneville, for reading and commenting on multiple chapters. I thank Sophie Goss and Jenna Zorik for reading all chapters multiple times and forcing me to articulate the story more clearly. Finally, I am grateful to my American MAGA friends and neighbors for many months of discussions that provided insight into their underlying thought processes and confirmed what I had learned from my teenage daughter: there is indeed often less to life than reason.

I The Rational Animal

Man is the only animal capable of reasoning, though many others possess the faculty of memory and instruction in common with him.

—Aristotle

There’s a logical explanation for everything, often mistaken for the reason it happened.

—Robert Breault

To ask questions about the role of reason in human affairs is, in the broadest sense, to ask questions about our place in the universe. What is the nature of man? Who and what are we? We have struggled with such questions for as long as we have been able to think about such things. Are we reasoning animals? Are we only reasoning animals? Is reason necessary? Is it sufficient? What ever happened to the animal passions? Have socialization and culture—constructions of the reasoning mind—allowed us to rise above them (like Katharine Hepburn’s character in the film The African Queen advocated [Huston, 1951]: Nature, Mr. Allnutt, is what we were put on this world to rise above), or do we need an account of human nature that reconciles the two? The reader will guess from the title of the volume that I make the case for the latter.

1 Food, Sex, Politics, and the Rational Animal

To proceed on this track, investigators would need to accept one grand but empirically robust premise—that higher aspects of the human mind are still strongly linked to the basic neuropsychological processes of lower animal minds.

—Jaak Panksepp

Much of life is about pursuing food, sex, and politics. Any adequate theory of human behavior must be able to explain these pursuits.

By far the most popular academic accounts of human behavior place the rational mind front and center (Cassirer, 1944; Durkheim, [1895] 2014; Simon, 1955). Humans bring the tools of reason to bear on these problems. Reason sets us apart from other animals. It allows us to successfully pursue not only food, sex, and politics but also art, science, and technology. This model is often referred to as the standard cognitive or social science reasoning model of human behavior (Tooby & Cosmides, 1995). After more than 20 years of trying to understand human decisions and choices just through the lens of reason, I have become skeptical of the explanatory scope of this standard model.

I’m convinced that reason is an integral part of who and what we are. I’m also convinced that, on its own, it is inadequate to explain much, if not most, real-world human behavior. It is only half the story. We do not have to look very far to understand what is missing. There is a commonsense model of behavior, embedded in the Western-Christian intellectual tradition, that recognizes not only reason but also animal passions (often characterized as the four Fs: feeding, fornicating, fighting, and fleeing) as determinants of human behavior. Our choices and decisions are a function of both. Not only is this much more intuitive, but we will see that the data demand such a model.

Despite common sense and data, such a model no longer gets serious consideration in large segments of modern society, including much of academia. I worry that the main reason is that many people, some academics included, hold variations on the meritless belief that humans no longer need to rely on instinct to survive, not when we have education, technology, and social norms (Pomeroy, 2011). The goal of this book is to push back against this widespread misconception, and articulate a commonsense model of human nature, called tethered rationality, that preserves the basic intuitive insight of the Western-Christian model—that both reasoning and nonreasoning systems are in play in human behavior—and can be discharged without divine intervention.

The animal passions, or nonreasoning behaviors in technical parlance, include autonomic behaviors, instinctive behaviors, and associative learning behaviors. These behaviors and their underlying mechanisms have been studied extensively over the past hundred years. They differ not only from reasoned behaviors but also from each other. They are hierarchically organized in terms of appearance on the evolutionary tree, are integrated, and are widely available across species, including humans. Humans also exhibit reasoning or rational behavior, which (I will argue) is unique to us. However, it does not supplant the evolutionarily older behaviors. Reason evolved on top of them, but it does not float untethered above them; it is tightly integrated with both bottom-up and top-down connections. This means that human behavior is a blended function of all these systems, not just reason (or any other individual system). Humans have a reasoning mind, but it is tethered to and modulated by evolutionarily older associative, instinctive, and autonomic minds.

I begin this chapter by introducing five examples of real-world decisions that are widely thought to be explained by reason. Before we can consider whether these examples are actually explained by models of reasoning, we need to introduce the notion of reason and rationality. This is initially done informally. With this preliminary understanding of reasoning in hand, I then evaluate each example to see if it can be explained just in terms of reason. I conclude that four of the five examples cannot be so explained. Satisfactory explanations for these require the introduction of evolutionarily older nonreasoning systems. A roadmap is then provided to foreshadow the argument for the model of tethered rationality and guide the reader through the subsequent chapters.

Examples of Reasoning in the Real World

Let’s begin by considering five real-world examples of reasoning and decision-making scenarios.

The first example is climate change, the ultimate existential issue of our time. The best science we have agrees that human activity is contributing to rising temperatures, which will reshape planetary weather patterns and geography and have detrimental, even catastrophic, effects on all life on earth. The scientific models could be wrong by either overestimating or underestimating the changes that will occur, but they provide the best information we currently have. Most governments and citizens accept the science and are willing to take some (limited) steps to mitigate the impact of human activity. However, the forty-fifth president of the United States, a number of US senators, and 40% of the American public believe that man-made global warming is the greatest hoax ever perpetrated on the American people (Revkin, 2003). They claim, without evidence, that the scientific models are incorrect. Even among the other half of Americans who do accept the science, there is considerable reluctance to undertake full remedial measures. This example illustrates two separate issues: that many people simply deny the science, without evidence to the contrary, and others seem to accept the science but fail to act on it. There seems to be a lack of rationality in both cases.

The second example involves weight management. Last year, I went to my doctor’s office for my annual checkup. After I stepped on the scale, my doctor advised me to lose 30 pounds. I agreed but complained that my busy schedule did not allow time to eat healthy meals and exercise regularly. My doctor replied, What fits your busy schedule better, eating healthy and exercising one hour a day or being dead 24 hours a day? Many of us have been in this situation, but few of us actually manage to follow our doctor’s advice. Notice that we do not question the doctor’s judgment. There seems to be considerable evidence linking obesity with the onset of various diseases (e.g., diabetes and heart disease) and premature mortality. Most of us do not have a death wish. Given that we want to live a long, healthy life, and given that we accept that obesity will impair and even shorten our lives, the rational, reasonable thing to do would be to lose weight. So, why don’t many of us comply with our doctor’s advice?

For our third example we turn to sex. In December 2006, John Edwards, a handsome, charismatic lawyer and politician, announced his candidacy for the 2008 Democratic nomination for president of the United States. He was among the frontrunners, along with Barack Obama and Hillary Clinton, for the nomination. In March 2007, it was revealed that his wife, Elizabeth, was suffering from stage IV breast cancer. Shortly thereafter, it came to light that he was having an affair with one of his campaign workers. In what world was this a rational choice? He was running for the highest office in the world, in a country that contains some of the most socially conservative, prudish, judgmental, evangelical voters. He must have known that if there was any hint of infidelity—even in the best of circumstances—his campaign was over. His circumstances were such that his wife was dying of cancer and receiving enormous emotional and moral support from the public. Any hint of infidelity in such circumstances would be suicidal. Evidence of the affair emerged in early 2008 and ended his candidacy overnight. How do we explain his choices?

The fourth example concerns healthcare, a topic that often comes up in discussions with my American friends. The conversations often take the following form:

Me: Given your very high premiums and the large deductible in your private healthcare plan, why don’t you support overhauling your healthcare system into a universal Canadian/European-type system whereby everyone can receive good equivalent healthcare at a lesser cost?

My American friend: Affordable healthcare would certainly be a great benefit to me. However, you see that guy over there? Yes, that one. He doesn’t work. He doesn’t pay taxes. He is a freeloader. If we had universal healthcare, he would get the same healthcare that I do, but he doesn’t deserve it. Therefore, I cannot support a universal system. (Another interesting response is the admission that, yes, that would probably be better than what we have, followed by passing shame and a disappointed sigh, but that would be socialism.)

My friend is willing to forgo a benefit for himself just so that someone undeserving does not receive an equivalent benefit. Again, it is hard to see the rationality in this choice.

For the fifth example we turn to a drug warning issued to doctors and patients by the UK Committee on Safety of Medicines in 1995. The warning stated that the third generation of birth control pills doubled (i.e., increased by 100%) the risk of life-threatening blood clots in the legs and/or lungs. Unsurprisingly, this caused great anxiety among women and resulted in a sharp increase in unwanted pregnancies and abortions in subsequent years. A closer examination of the study showed that for every 7,000 women who took the second-generation pill, one developed thrombosis. By contrast, for every 7,000 women who took the third-generation pill, two developed thrombosis. So, while the relative risk did increase by 100% as advertised, the absolute risk was an increase of 1 in 7,000 women (Gigerenzer, 2015). This hardly seems to warrant the panic that ensued, so how can we explain it?

These are five (very different) examples of everyday, real-world decisions or choices. Other examples will be introduced throughout the book. Even though I have not yet formally introduced the idea of rationality, I’m confident most readers will agree that each example illustrates a choice that seems less than fully rational. I will not go so far as to say that they are irrational. In the cases of examples one through four, I will argue that they are arational—that is, they involve noncognitive factors.

The most popular academic models that we have for explaining these behaviors are the cognitive reasoning and decision-making models, buttressed by distinctions between analytic and heuristic reasoning, such as the fast and slow thinking model popularized by Daniel Kahneman (2012), or by notions of motivated reasoning (Kunda, 1990) or even sloppy reasoning (Pennycook & Rand, 2019). Such models will be introduced and considered in chapters 7 and 13. They provide satisfactory explanations for a number of phenomena, including example five, but lack the requisite machinery to deal convincingly with examples one through four, which are the ones of interest in this book.

To make sure we are all on the same page, I offer an initial introduction to the notion of rationality and decision-making and then return to address the preceding examples.

What Is Rationality?

Man is widely considered to be the reasoning or rational animal. But what does this mean? To invoke reason or rationality is to say that human behaviors or actions are explained by postulating beliefs and desires and a principle of coherence that guides our pursuit of the latter in the context of the former. By coherence I mean roughly making sense. Coherence is a relationship that holds between thoughts, propositions, or sentences. In the first instance, it is a basic, primitive, intuitive notion, though it can be considerably enhanced with education. For example, if I believe that all Americans are intelligent, and all Fox News viewers are American, then it would be coherent or reasonable for me to infer that all Fox News viewers are intelligent. Given the same beliefs, it would not be coherent to infer that no Fox News viewers are intelligent. This example illustrates a particularly extreme case of coherence found in deductive arguments, referred to as validity, where the truth of the given information (or beliefs) is sufficient to guarantee the truth of the conclusion, but it is worth noting that validity does not evaluate the veracity of the premises that all Americans are intelligent and all Fox News viewers are Americans; it merely determines whether a conclusion follows from or is entailed by them. We can consider validity as coherence in the narrow sense of the term and additionally have a broader sense of the term, corresponding to soundness in logic, that also takes into consideration the veracity of the premises. In this broader use of the term, we would step back and evaluate (and either accept or reject) the truth of the premises before drawing the inference.

On a recent trip to New Delhi, India, one afternoon I observed Indian fruit bats dangling from tree branches like so many brown and black cloth sacks. Based on this observation, I formulated the belief that Indian fruit bats spend the afternoon dangling from tree branches. This is a plausible or coherent inference based on my observations, but notice that it lacks the certainty of the preceding inference about Fox News viewers. Further observations (or consultation with bat experts) might reveal that this behavior is a peculiar habit of fruit bats in this particular region of India. In this case, I would have to modify my belief for it to be consistent with the facts in the world. Absent additional information, it is coherent for me to believe that Indian fruit bats spend afternoons dangling from tree branches. Given the same evidence, it would be incoherent for me to conclude that Indian fruit bats do not spend the afternoon dangling from tree branches or spend the afternoons diving for crayfish in shallow rivers.

Coherence relations between premise and conclusion are disrupted by inconsistency, indeterminacy, or irrelevance. Inconsistency is illustrated where the conclusion No Fox News viewers are intelligent is drawn from the beliefs that All Americans are intelligent and All Fox News viewers are Americans. An example of indeterminacy occurs if I tell you Mary is taller than George, and Mary is taller than Michael and ask you the height relationship between George and Michael. The premises do not provide sufficient information to draw any inferences about the relative heights of George and Michael. An example of failure of coherence through irrelevance would occur if, given the belief that all Americans are intelligent and the belief that Indian fruit bats spend afternoons dangling from tree branches, I conclude that global warming is caused by human activity. In this case, the issue of coherency does not even arise, because the three propositions are unrelated.

From Rationality to Decision Theory

Reasoning is about maintaining coherence in belief networks. Life is about actions. Reason mediates action by determining choices consistent with specific goals, given specific beliefs. Choice selection is studied by decision theory. We get from reasoning to decision-making by overlaying some model of human goals on top of the model of rationality. These models are usually based on maximizing self-interest. A historically popular one is the Homo economicus model. In this account, man is intrinsically a self-interested utility maximizer as a consumer and a self-interested profit maximizer as a producer.¹ These become the goals of the individual. Rational actions are those that are expected to advance goals in light of beliefs.

I will illustrate this standard model of decision-making with the controversial US decision to invade Iraq in March 2003. While I have no privileged access to the particulars of the decision-making process, its overall form would be something like that depicted in figure 1.1. It would begin with a goal or desire that needs to be achieved, such as securing the Iraqi oil leases. This goal would be explored or expanded via subgoals. One subgoal option might be negotiation. Another might be to take the oil by force if certain conditions can be met, such as: assurance of success, clean surgical intervention and withdrawal, that the value of the oil leases be greater than the cost of the invasion, and that Iraq be able to pay for its own reconstruction costs. In this example, these conditions are believed to be met (and negotiation is not considered feasible or cost-effective), leading to the subgoal of invading Iraq. However, there are accompanying beliefs that suggest most Americans (and the world community) will not support an unprovoked invasion, even if it means access to cheap oil. This results in another subgoal to pause and reconsider. There are accompanying beliefs that most Americans (and the world community) would support a defensive war against a tyrant. This generates another subgoal of launching a campaign to vilify Saddam Hussein and convince Americans that Iraq has weapons of mass destruction that are an imminent threat to the United States (which has more weapons of mass destruction than all other countries combined) and its allies. It is determined that the propaganda campaign is successful and there is sufficient support within the country for the invasion. Given all this, the rational decision is to invade Iraq; each step follows coherently from the previous goal or subgoal plus beliefs.

Figure 1.1

An example of the rational mind at work using a hypothetical reconstruction of the US decision to invade Iraq in 2003. Each subgoal follows coherently from the preceding goal or subgoal plus beliefs, eventually resulting in an action. The integration of goals and subgoals plus beliefs via the coherence relation is the nexus of the reasoning step.

However, this model is an oversimplification. It assumes that the beliefs or information at hand are complete and certain. But how certain are we that Iraq can repay its own reconstruction costs? 100%? 10%? 73%? Are there any constraints on the desire to take the oil by force? If the financial cost of the war equals or exceeds the benefits of the oil, do we still want to pursue this desire? In real-world situations, information is always incomplete and uncertain, and even the relative utility of different desires cannot be confidently ascertained and ordered. These complications transform the problem of inferential coherence from the realm of logic to the realm of probability theory (see figure 1.2). Coherency is then determined by applying the probability calculus to the model. The rational choice is the one with the highest utility value. One consequence of this shift is that the criterion of coherence morphs to an optimality criterion. However, for our current purposes, these complications are not material. It is still coherent to select the option with the highest expected utility (see figure 1.2). I have chosen to use the concept of coherence rather than utility as central to rational decision-making throughout the book.

Figure 1.2

Simple decision tree and utility function. One might model the decision to invade or make a deal as follows. The chances of a successful invasion are 0.75, while the chances of failure are 0.25. The chances of a successful negotiation are 0.15, while the chances of failure are 0.85. The value assigned to both the successful invasion and successful negotiation is +100. The value assigned to a failed invasion and failed negotiation is −100. Based on these values, the utility of invasion is +0.5 and the utility of negotiation is −0.7 (utility = Σ(probabilityoutcome × valueoutcome)). Notice that decision theory provides no guidelines for assigning probabilities of outcomes and the value of the outcomes, but once these numbers are (magically) assigned, simple probability theory allows us to coherently calculate expected utility. The rational choice is the one with the highest expected utility.

This example is offered as a simplified illustration of the machinery of standard decision-making models. There are two points worth noting. First, the postinvasion justification (when no weapons of mass destruction were found)—that American lives and resources were expended so the Iraqi people could benefit from regime change and democracy—is irrational because it violates the basic tenets of maximizing self-interest. Second, I’m not claiming that this rational model is sufficient to explain the invasion of Iraq. On the contrary, I’m certain that a number of nonrational factors considered in this book were significant factors in making the decision. More generally, I’m claiming that such standard models of rationality cannot adequately account for much of human behavior, including the invasion of Iraq and four of the five examples introduced earlier. Understanding this claim requires reviewing each example more closely, beginning with global warming.

Rationality in the Real World: Global Warming Example

The basic questions around climate change are Is the earth warming? and Is human industrial activity contributing to it? Most scientists answer yes to both questions (The National Academy of Sciences & The Royal Society, 2020). Many members of the public agree, but at least 40% of Americans vehemently disagree. The same data are available to all. We are all rational, so why the discrepancy in opinion? Let us consider the argument and the various sources of dissent to see rationality working, failing, and being irrelevant.

The argument climate scientists make for man-made climate change is summarized as follows by the National Aeronautics and Space Administration (2020): data indicate global temperatures have been steadily rising since the 1800s (the start of the Industrial Revolution), resulting in melting of the polar ice caps and rising sea levels. There can be many natural sources for temperature increases, such as variation in solar activity, volcanic activity, and even slight shifts in Earth’s trajectory around the sun, and these have indeed resulted in past climatic changes. But the timescale and fingerprint of the changes we are currently experiencing are not consistent with any of these natural causes. Examination of ice cores from Antarctica reveals that carbon dioxide levels have been relatively stable throughout the past 800,000 years but have shot up dramatically over the past hundred years. When we incorporate the data about excess introduction of carbon dioxide into the atmosphere as a result of human fossil fuel activity and disruption of the natural carbon-oxygen cycles, the projected greenhouse effect is very similar to what we are actually experiencing. Therefore, it is reasonable to believe that human activity (such as carbon dioxide emissions) is a large causal factor in global warming.

This conclusion is plausible, perhaps even compelling, but not certain. One can probe, question, and doubt. Let’s examine some possible reasons for rejecting the argument offered by nonbelievers by reviewing a question-and-answer session on climate change, held in June 2010 at the University of New South Wales, called The Sceptics (2010). It was moderated by Jenny Brockie and featured climate scientist Professor Stephen Schneider from Stanford University and some ardent skeptics from the Australian general public. The first skeptic questioned by the moderator was Tania.

Moderator: Tania, do you believe in man-made climate change?

Tania: Man-made? Not at all.

Moderator: Why?

Tania: No one has proven to me that it’s man-made at all. What I say is it’s a big hysteria just for money. The only reason you’re getting grant money is because of climate change. The planet is warming is the only reason you’re getting grant money. If we didn’t have this hysteria there would be no grants. There would be no people making money at all.

In this case, the argument for climate change is not actually in play. Tania’s objection does not consider the relation between the evidence and the conclusion. Tania is attributing a disingenuous or malicious motive to climate scientists and is offering an ad hominem response. Scientists are simply lying to pad their pockets with grant money. This objection is a case of disagreeing with a conclusion but for reasons that have nothing to do with the coherence of the argument. Many real-world disagreements fall into this category. A similar technique can be used to endorse arguments that are offered by friends and people that one admires. The argument itself does not matter. Coherence relations between evidence and conclusion are not in play. Therefore, such objections (or endorsements) do not belong to the realm of the rational. One might think that educating Tania about the individualistic and competitive nature of science and scientific grant funding may dissuade her from her misconception, but as we will see in chapter 13, it probably