Review - Why We Lie
The Evolutionary Roots of Deception and the Unconscious Mind
by David Livingstone Smith
St. Martin's Press, 2004
Review by Alex Sager
Jul 30th 2004 (Volume 8, Issue 31)

David Livingston Smith's Why We Lie: The Evolutionary Roots of Deception and the Unconscious Mind follows in the tradition of Steven Pinker's How the Mind Works, Matt Ridley's The Origins of Virtue and Robert Wright's The Moral Animal. Like those books, Why We Lie is well-written and likely to be embraced by fans of evolutionary psychology (as the blurbs on the back of the hardcover suggest). Readers of these works will find much of his material familiar. Unfortunately, for those who have a sympathetic, but more skeptical view towards evolutionary psychology, it will seem a wildly speculative and generally unsatisfactory mishandling of a potentially fascinating topic.

The book's central idea is that we possess a "Machiavellian module" that enables us to deceive others and detect deception. This module takes its name from the Machiavellian intelligence hypothesis popularized by the Richard Byrne and Andrew Whiten's anthology Machiavellian Intelligence. The Machiavellian intelligence hypothesis draws on research on non-human primates' social relationships, which often involve deception. It speculates that brain size and high intelligence are caused by the selective pressures of social competition, rather than the more traditional belief that it came from tool use and other aspects of the physical environment. If this is true, it is quite possible that our minds have adapted to commonly reoccurring social problems faced by our Pleistocene ancestors. According to Smith, this module operates unconsciously, with the paradoxical consequence that we often deceive ourselves about our deceptions.

What are we to make of these claims? The notion that we have specialized mechanisms for social intelligence is a common idea in evolutionary biology, though our knowledge is currently highly speculative. The modular theory of mind, which states that the human mind is composed of a large number of domain-specific modules, also seems to me quite plausible, in some form or other. It is reasonable to believe that there have been selective pressures to both deceive others and detect lies, possibly resulting in one or more specialized mechanisms devoted to this task  (There are serious questions about the nature of modules, their relation, if any, to domain-general processing devices, their specific domains, etc.  Since Smith wisely refrains from specific claims about the architecture of his Machiavellian module, we can ignore this). Similarly, the fact that much of the mind operates unconsciously is uncontroversial (indeed, if anything needs explaining these days, it's conscious thought). The existence of unconscious deception, even without the support of research in the social sciences, seems likely.

Still, what possible advantage could self-deception have? One possibility suggested by Robert Trivers is that self-deception is an adaptive mechanism: conscious deception often causes distress, partly because of the fear that the deception will be detected. For this reason, deceivers may not be able to control subtle tics that alert their victims. But if they, themselves, were unaware that they were deceiving others, this problem would be eliminated. Self-deception, then, becomes a tool for more effective manipulation.

            I'm willing to seriously consider this hypothesis, but it raises questions. An obvious one is why selective pressures didn't simply create better conscious deceivers. Since conmen (and women) and expert poker players seem highly adept at concealing their lies, this hardly seems impossible. Smith asserts that "if human beings had the knack of consciously making penetrating inferences about each other's motives and strategies, our insights would come at a high price (146)." This price would be losing the benefits of unconscious self-deception, since Smith seems to think that folk-psychology must treat oneself and others equally. This seems to me to be both false and highly speculative. First, many people do seem to be quite good at consciously interpreting people's behavior (it's something that can be improved through practice). Applying this knowledge to one's self may actually be an advantage. Second, why is it so implausible that our folk psychology could treat our self-conception and our conception of others differently? Can't a person with a grossly inflated self-opinion also make brutally accurate assessments of others? Smith's claim needs to be argued.

             The problem is that induction from natural selection is rather unreliable. Natural selection depends on previous adaptations, lucky mutations and the physical and social environment. Just because an explanation sounds plausible, doesn't mean it's correct. Evolutionary psychology is notoriously vulnerable to "just-so stories," explanations that conveniently invoke selection pressures to explain all sorts of phenomena, with little independent support. The value of evolutionary psychology is that it provides a means for developing hypotheses that can then be experimentally tested. Smith doesn't give much evidence to accept Triver's suggestion.

Instead, he takes for granted the advantages of unconscious manipulation and turns to psychoanalysis to reveal the nature of this Machiavellian module through the analysis and interpretation of language. Smith's idea is that our unconscious mind is constantly on the lookout for conflicting interests, deception and manipulation. He suggests that our speech is likely to contain "unconsciously coded messages" in "situations involving covert conflicts of interest between the speaker and some other person(s), and in which it would be disadvantageous to speak openly about these conflicts of interest (128). (Smith's italics)" But why would this unconscious Machiavellian module be represented in verbal behavior? These codes, if they indeed exist, seem to be some sort of by-product that, for some strange reason, gets expressed.

Let's look at his examples, accepting his interpretations, for the moment, as correct. In one example, Smith mentions attending a faculty event with his wife, a much younger woman of color. One of his colleagues brought up out of the blue the fact that his cousin had recently adopted a child from Africa and that he was "too old for that sort of thing (129-130)." Smith's explanation is that we are witnessing the Machiavellian module in action. But while there may very well be unconscious forces at play here (ruling out the possibility that the colleague was simply being nasty), I'm not sure I understand what this has to do with conflicting interests, deception or manipulation. If anything, the colleague is unconsciously revealing his prejudices, exactly the opposite we'd expect from a well-functioning Machiavellian module!

            Another example involves a young woman who, on Smith's interpretation, compares absent classmates (it was an early morning class on a cold day) to a man who went hunting, letting his three-year-old child freeze to death in the truck (129). Once again, while the absent students may have violated an implicit agreement, this doesn't seem to have much to do with deception. These two examples are typical: somebody is expressing something they'd perhaps like to say, but can't because of social norms concerning appropriate speech. Perhaps these norms are so strong that we aren't even consciously aware of this desire (though this isn't obviously the case). For some reason, the desire to communicate is stronger than the need to keep quiet and the message emerges in a disguised form.

            Smith suggests later on in the book that "the adaptive function of encoded communication is to warn one's allies of the presence of a social predator (171)." But I'm not sure this makes sense. If there's a need to code communication, then we must be in the presence of threatening rivals. But why would allies be more adept than enemies at unraveling these unconscious communications? Smith's account simply isn't compelling.

            I suspect that there are psychological mechanisms for deceiving and detecting deceptions. I also accept the strong possibility that these mechanisms operate largely unconsciously. Finally, I would be surprised if they weren't the product of selective pressures coming from the social environment. The problem is that Why We Lie doesn't have much to offer these three plausible hypotheses. It's rife with speculation that doesn't consider the obvious counterarguments, considerably weakening its case. We're left with a few interesting suggestions and little else.

© 2004 Alex Sager

Alex Sager is a doctoral candidate at L'Université de Montréal working on the ethical implications of biotechnology and public policy.

David Livingstone Smith responds to Alex Sager's review of Why We Lie

I am grateful to have this opportunity to respond to Alex Sager's highly critical review of my book Why We Lie: The Evolutionary Roots of Deception and the Unconscious Mind.  I don't deny that the book has its faults, and contains much that readers might want to take issue with, but I think that Sager presents a misleading picture. I will confine myself to six specific points that cry out for clarification most urgently.

First, Sager characterizes the book as "wildly speculative" and "rife with speculation".  Anyone reading his remarks might conclude that I've attempted to palm off wild and woolly conjecture as sober fact.  In fact, the Why We Lie is avowedly and necessarily speculative, because it deals with topics that have received very little empirical scrutiny. I underscore this point repeatedly, some have said excessively, throughout the book beginning with the Preface, from which I quote:

A book purporting to be scientific, but which builds arguments on what may seem to be the flimsiest of foundations, is bound to disappoint some readers.  Where is the experimental data?  I plead guilty of not having provided adequate empirical support for the distinctive views advanced in this book.  Although immensely powerful and valuable, experimental research is not the be-all and end-all of cognitive science. Science is the disciplined passion to find out, to make sense of the puzzles presented by the world around us and within us, and to make reasoned extrapolations from what we already know, or at least think that we know.  The book points the way to empirical research into self-deception and unconscious communication.  It shows scientists of the mind where they might look, and just what phenomena they ought to look for. You might say that this book resides on the borderland between science-yet-to-be, in a wild place where there are few paths and the signposts are scarce and difficult to read (p. 4).

Second, in connection with Trivers' theory of self-deception, Sager asks why natural selection did not turn us into better conscious deceivers instead of self-deceivers.  I had thought that the most plausible answer to this question was implicit in my discussion as well as in Trivers' classic accounts: conscious deception is risky, costly and relatively inefficient.  All conscious mental states are potentially public, and it can be hard work making sure that the cat never gets out of the bag, especially when engaged in sustained social interaction with the persons whom one is attempting to deceive.  Master poker players, con-artists and other virtuosos of conscious deception have to be intensely vigilant to pull off their feats of mendacity.  Unconscious deception is a lot easier and probably much more efficient.

Third, Sager states that I seem to think that 'folk psychology must treat oneself and others equally.' I never made such a silly claim.  In fact, in Appendix II of Why We Lie presents a list of the ways in which we skew our perceptions, memories and reasoning in self-serving ways.   My argument turns the idea, commonplace in the philosophy of mind, that we make sense both of ourselves and of others by applying a single interpretative framework and not on the notion that we apply this framework in precisely the same way to both self and other.

Fourth, Sager claims that I 'turn to psychoanalysis' to understand the nature of unconscious Machiavellian intelligence.  Given the contemporary consensus that psychoanalysis is not scientifically credible, this makes the book sound rather wacky. Although the general verdict on psychoanalysis expressed in Why We Lie is explicitly negative, I do not regard all things Freudian as prima facie worthless. I give Freud his due as a pioneer of the neuroscientific theory of mind and probably the first architect of a detailed theory of unconscious cognition.  I also call attention to a few remarks in the psychoanalytic literature that may provide valuable clues about how unconscious social intelligence operates.  Freud hinted in 1913 that the human brain might be endowed with a cognitive module specifically adapted for making inferences about others' concealed mental states and his Hungarian colleague Sándor Ferenczi suggested, almost two decades later, suggested that such unconscious inferences are expressed in verbal disguise. The latter part of the book develops these ideas in an evolutionary biological context and in a more contemporary scientific idiom.

Fifth, Sager states that he does not understand how my vignettes purporting to illustrate unconscious communication cohere with my larger theoretical claims. To give just one example, I describe introducing my youthful and extremely attractive Caribbean wife to an academic colleague of roughly my own age.  The colleague then remarks, seemingly out of the blue, that his cousin adopted a child from Africa and was 'too old for that sort of thing'. In the book, I suggest that the remark about a child from Africa was an unconscious portrayal of my Black wife, and that the cousin who was 'too old' represented me.  Sager remarks "I'm not sure I understand what this has to do with conflicting interests, deception or manipulation."  In fact, I claim only that the presence of covert conflicting interests drives this phenomenon.  Deception and manipulation are often present, but they are not necessary.  I assumed – apparently wrongly – that there was an obvious connection between this particular interpretation and my more general theoretical position.  Spelling it out, the driving force behind my colleague's remarks may have been covert sexual rivalry concerning a reproductively valuable female.  In unconsciously presenting my youthful wife as a 'child' and myself as 'too old' he disparaged the relationship.  This was all concealed in a seemingly innocuous and manifestly irrelevant story about one of his relatives.

Finally, Sager is unhappy with my claim that the adaptive function of these veiled messages is to alert allies to the presence of a social deviant.  He argues that because deviant individuals would also be able to decode unconscious meanings, the disguise would serve no real purpose. This completely misses the point.  If I am in a crowd and somebody steals my wallet I might yell "Stop, thief!'   I don't care that the thief understands what I'm saying, because the purpose of yelling is to alert the crowd.  By the same token, it doesn't matter that the person tagged as a social predator understands the meaning of an unconscious message, because, I suggest, the point of the message is to galvanize other members of the community.  I don't think we unconsciously encode Machiavellian messages prevent others from understanding them. I think it's far more likely that we encode them primarily to preserve the fog of self-deception that envelops our social lives but which, paradoxically, helps us to promote our interests by deceiving and manipulating others.

© 2004 David Livingstone Smith

Smith complains that I ignore his repeated admissions that his book is "avowedly and necessarily speculative". This wasn't my intention. Speculation, even speculation from "a wild place where there are few paths and the signposts are scarce and difficult to read" is, in my opinion, a valuable and entirely respectable activity. Smith is right not to limit scientific to empirical research. Still, it is important, when searching for imaginative solutions, to have a clear idea of how your ideas might be supported by empirical research. There is a crucial difference between mere speculation and speculation that can genuinely contribute to what we know. I did not mean to criticize Why We Lie simply for being speculative; rather, I had doubts about the plausibility of its speculations and whether or not they could eventually lead to anything more.

I will try to make this clearer with an example. When reading Why We Lie, I was surprised to see no reference to Lena Cosmides and John Tooby's justly famous article "Cognitive Adaptations for Social Exchange", which can be found in the anthology The Adapted Mind. In this article, Cosmides and Tooby point to the well-known fact that even people with a background in logic and science do surprisingly bad on relatively simply problems like the Wason selection task. In Wason selection tasks, subjects asked to falsify a conditional hypothesis of the form If P then Q, using a set of four cards, with letters and numbers written on each side. For example, the cards could be A, D, 3 and 4, and the conditional might read: "If the card has a vowel on one side, then it has an odd number on the other." Subjects are invited to turn over two cards and usually choose the A and the 3. The correct answer is the A and the 4; the 3 is irrelevant since the conditional doesn't say that every odd number must have a vowel on the back. A 4 card with a vowel on the other side, on the other hand, would effectively falsify the conditional.

What is interesting about this task is that people's scores are significantly higher when you substitute the abstract content and present four cards representing rules from social contracts. To use an example cited by Cosmides and Tooby, the cards might read "drinking beer", "drinking coke", "25 years old" and "16 years old" and the conditional could be "If there person is drinking beer, she must be over 21." In this case, it is obvious that the correct cards are "drinking beer" and "16 years old". To explain the fact that people do much better with these examples, despite the fact the abstract reasoning is identical, Cosmides and Tooby employ evolutionary reasoning. They suggest that one of the fundamental problems encountered by our hunter-gatherer ancestors is cheating, people who violate social contracts. This suggests that there may have been selective pressures that rewarded individuals who were capable of detecting cheating. This leads them to posit that we have evolved a "cheater-detector module" especially designed to foil contract violators.

I don't want to dwell on this example or necessarily endorse it, but rather to distinguish it from much of Smith's book. In my opinion, this cheater-detector module is a perfectly respectable scientific hypothesis based on evolutionary reasoning. It is supported by empirical experiments on the Wason selection task and Cosmides and Tooby attempt to support their hypothesis by discussing competing interpretations of this evidence. We can also see how further experiments and evidence could serve to confirm or refute it.

Contrast Smith's use of Triver's account of deception. Like Cosmides and Tooby's example, it uses evolutionary reasoning, but it's often difficult to decide how his hypotheses could be developed or confirmed. When I asked why evolution didn't simply create better conscious deceivers I was making a point about evolutionary reasoning: it always needs something else, a background of evidence, experiments and theories which limit the scope of interpretation. Smith may be right that "conscious deception is risky, costly and relatively inefficient", but he should also realize that these claims are absolutely fundamental to his entire argument and need further support. How should we characterize deception in general? How risky, costly and inefficient is conscious deception? In what ways? In what domains? Are there essential differences between conscious and unconscious deception? What evidence can be brought to bear on these questions? What experiments could we conduct to further our understanding? I realize that Smith is working in a relatively novel domain and that it would be unfair to hold him to an unreasonable high standard, but this is the sort of work that needs to be done. The point is this: if we are simply relying on evolutionary speculation to support our theory, we're on very flimsy ground.

Third, Smith takes exception to my claim that he "seems to think that folk-psychology must treat oneself and others equally". Perhaps I wasn't sufficiently clear, but it was directed at his claim that social inferences of oneself and of others must take place within the same framework. I was particularly struck by a particular passage which I will now quote in length:

Mother Nature has seen to it that the conscious mind is relatively blind to the nuances of social behavior. It is easy to understand why this turned out the way it did. If human beings had the knack of consciously making penetrating inferences about each other's motives and strategies, our insights would come at a high price. Self-deception would become much more difficult, and this would rob us of its vital benefits. To understand why, consider a physiological analogy. It is impossible for a person to damage his or her eyes in such a way as to make the unable to see only certain kinds of objects. […] If one is blind, one loses a whole dimension of experience. The same principle applies to the social "blindness of the conscious mind, which provides us with relatively impoverished portrayals of both our own actions and motives and those of others. All social inferences flow from a common set of assumptions, an informal folk-psychological theory of human nature. If the theory is biased, it will deliver faulty appraisals of everyone: not only of oneself, but also of other people. […] The knife of self-deception cuts two ways: you cannot maintain a highly distorted conception of yourself side by side with a true estimate of others.

To begin with, the physiological analogy is purely rhetorical. There is no reason to think that any principles taken from physical blindness also apply to social blindness. Leaving that aside, why is it that all "social inferences flow from a common set of assumptions"? I question Smith's claim, despite the fact it is "commonplace in the philosophy of mind", that we use a single interpretative framework. Once again, where is the evidence that supports this? I'm not willing to rely on the authority of philosophy of mind. Smith himself may in fact contradict it by drawing attention to the fact that "we skew our perceptions, memories and reasoning in self-serving ways". If this is the case, why is it necessarily the case that we use the same interpretative framework? Couldn't it be possible that we use two (or more) interpretative frameworks? After all, how exactly do we count interpretative frameworks? I don't pretend to know how to resolve this problem, but believe that agnosticism is the correct stance. Once again, we need more evidence.

Fourth, I hadn't meant to dismiss psychoanalysis or Freud. Like Smith, I believe both contain valuable insights. Unlike Smith, I am extremely skeptical that "unconscious inferences [of a cognitive module adapted for making inferences about others' concealed mental states] are expressed in verbal disguise". Nothing Smith has said convinces me that the analysis of speech is a valuable tool for revealing the unconscious mind. What bothers me is I don't see how, once we have a number of plausible sounding interpretations, we can decide between them. In some ways, I think the history of psychoanalysis shows this: the "correct" interpretation often seems to have more to do with the particular theorist's idiosyncrasies than any really convincing theory. The result is something closer to literary criticism, something which may be of therapeutic value, but is unlikely to increase our knowledge of the mind. I don't see how Smith's placing "these ideas in an evolutionary biological context and in a more contemporary scientific idiom" limits plausible interpretations. The example of Smith's colleague's story about his cousin being too old to adopt a child from Africa may have undertones of sexual rivalry, etc., but is this obvious? Is the remark necessarily unconscious? Aren't there other competing interpretations? Couldn't his colleague simply having been (misguidedly) attempting to demonstrate affiliation? Isn't it possible that his remark was completely coincidental? Finally, how are we going to resolve these questions? I can imagine different people having competing intuitions and I'm not sure how we could determine which ones are the best.

Finally, I'm not sure Smith makes clear the adaptive value of veiled messages that point out social deviants. Smith claims that this messages work in the same way that he might yell, "Stop thief!" to alert the crowd that someone has stolen his wallet. But is it really unimportant that the culprit social predator understands these veiled messages? Smith suggests that these messages serve simply to "galvanize other members of the community." But I'm puzzled why we would evolve complex and most likely costly mechanisms for sending and decoding messages if not to conceal them. Smith believes it has to do with preserving "the fog of self-deception that envelops our social lives but which, paradoxically, helps us to promote our interests by deceiving and manipulating others". As my remarks above show, I'm still not convinced that this follows.

© 2004 Alex Sager