Review - Brain Fiction
Self-Deception and the Riddle of Confabulation
by William Hirstein
MIT Press, 2005
Review by Neil Levy, Ph.D.
Mar 9th 2005 (Volume 9, Issue 10)

Philosophers who are engaged in the project of attempting to understand the human mind increasingly turn to the neurosciences for inspiration and information. Hirstein is one the growing number of such philosophers who are at much at home in the neurosciences as in their home discipline. He has himself engaged in experimental work, and published several articles with neuroscience luminary V.S Ramachandran. In this book, he attempts to draw philosophy and neuroscience together, in order to contribute to understanding the mind and some of its pathologies.

The pathology, or pathologies, with which he is concerned is confabulation. Clinically, 'confabulation' is normally used to describe the manufacture of sometimes plausible, sometimes wildly implausible, fictions, especially by sufferers of Korsakoff's syndrome. Korsakoff's syndrome is characterized by severe amnesia, affecting autobiographical memory. Sufferers are unable to form new, lasting, memories of events.  They are therefore typically at a loss to know what they are doing in a hospital. When they are asked where they are, they confabulate: typically, they respond to cues in their immediate environment to construct a story. They might, for instance, take the doctor's white coat to be a baker's apron, and, filling in some detail from their preserved memory of earlier life, confabulate that they are visiting a friend who is a baker. Asked what they did yesterday, they might recount an event that happened to them forty years earlier. Confabulation is, apparently, not lying: confabulators appear to sincerely believe the tales they concoct.

There are a number of other strange neurological syndromes that centrally or typically involve the assertion of falsehoods without intentional lying. They include anosognosia (denial of illness, typically, the denial that one is suffering from hemiplegia following a stroke), Anton's syndrome (denial of blindness), and delusions such as Capgras's delusion (the delusion that people to whom the patient is close have been replaced by impostors) and Cotard's delusion (the delusion that the patient is dead). Hirstein argues that all of these syndromes should be understood as confabulatory syndromes, and given a unified explanation. He also holds that a unified account of confabulation will shed light on several other disorders, most notably sociopathy and obsessive-compulsive disorder.

All of the confabulatory syndromes, as Hirstein recognizes, appear to involve brain-damage. Examining which areas of the brain are compromised therefore seems to be a promising approach to explaining them. After an extended (and, for a philosopher, somewhat tedious) excursion through neuroanatomy, Hirstein advances a two-stage explanation of confabulation. He suggests that normal brains have separate modules (my term, not his; no implications about encapsulation are intended) for the generation of explanatory hypotheses and for checking these hypotheses for consistency and plausibility. Confabulation occurs when both modules are damaged. In that case, wild hypotheses are produced by the generation module, and are allowed to pass (or go unexamined) by the checking module. When the damage to one or the other module (or both; Hirstein is far from clear on this point) is domain-specific, we get confabulation restricted to a certain topic. We get, for instance, the patient who remains globally rational, but denies that her left arm is paralyzed. When the damage is more widespread, the confabulation spreads across more domains.

Hirstein hopes that by understanding the ways in which the mind breaks down, we can get a better understanding of its normal processes. Specifically, he hopes that understanding confabulation will shed light on normal self-deception. He advances some intriguing hypotheses about how self-deception, which he understands in a relatively traditional way, could be possible. Self-deception, understood in a traditional fashion (rather than in the deflationary way proposed independently by Al Mele and Annette Barnes) is puzzling, because it requires that otherwise rational agents are, not merely mistaken about some proposition, but actively deceiving themselves about it. Do we not need to know the truth about something in order to deceive ourselves about it? And if we know the truth, then are we not undeceived? How is it possible for someone both to believe something and not to believe it?

Hirstein suggests that the same mechanisms that operate in confabulation may be at work in self-deception. In confabulation, the checking mechanisms are damaged or destroyed. In normal self-deception, they are functioning normally, but there may be brain processes that direct the self-deceptive representation away from the checking processes. He also suggests an intriguing way in which a self-deceptive person might be able to, in some sense, simultaneously maintain contradictory beliefs. Hirstein suggests that full-blown beliefs are represented conceptually in the mind, but that there are inferior grades of representation. For instance, information might be represented in what he calls analog form, in which it is not conceptualized. A topographic map of space or of one's own body is one kind of analog representation. Though we may not be able simultaneously to maintain conflicting (conceptual) beliefs, we can maintain a conceptual belief which conflicts with one or more of our analog representations.

How do these two factors allegedly at work in self-deception interact? Is either sufficient for self-deception, or are they jointly necessary? Hirstein does not say. Unfortunately, this kind of lack of clarity about the mechanisms he proposes is typical. Consider his explanation of his main target, confabulation. Hirstein suggests that mind-reading deficits – an inability to represent or to simulate the mental states of others' – may be important in confabulation. But how? When he advances his official two-stage theory, mind-reading is dropped. And indeed, it is difficult to see how a mind-reading deficit could be integrated into his theory. One possibility is that we might utilize the responses of others as an additional check on our hypotheses; so that the orbitofrontal cortexes of interlocutors could replace or supplement our own. If this were the case, we might get short-lived confabulatory syndromes, in patients whose orbitofrontal cortexes are damaged but who retain sufficient mind-reading facility to compensate. Equally, however, it may be that confabulators do pick up on the responses of others, but this only spurs them to ever-changing stories; some case studies seem to provide evidence for this hypothesis.

In any case, the two-stage theory seems to have a stage too many. Hirstein suggests that confabulation requires both the generation of implausible hypotheses, and the failure of checking mechanisms. This is clearly false for some confabulations, and may be false for all of them. Some confabulations are, as he notes, quite plausible: if you did not know that the confabulator had been hospitalized for the past six months, you would have no difficulty accepting their stories. Why think, then, that the mechanism for generating explanations needs to malfunction to produce such confabulations? Indeed, it may be that even the wildest hypotheses are generated by normally functioning mechanisms. It may that fantastic scenarios are routinely generated by sub-personal mechanisms in the brain of normal persons. In that case, all confabulations might simply be failures of checking mechanisms. Confabulations might be verbal analogues of utilization behavior: we may regularly produce such hypotheses, and would accept them but for the winnowing effects of our checking mechanisms. Indeed, Hirstein himself seems to suggest as much at various places in the book. When he advances his official explanation, however, the suggestion is dropped.

Should we substitute a one-stage explanation of confabulation for the two-stage mechanism Hirstein proposes? Not, I suggest, if our task is to explain the range of phenomena that Hirstein calls confabulation. These phenomena seem to be too disparate to be given a unified explanation. It may be that neither the generation nor the checking mechanisms are damaged in some confabulators. Some confabulation may be 'gap-filling', of an entirely normal kind. Perhaps we routinely explain our actions, to ourselves and to others, by reference to environmental cues (this would be a type of extended mind hypothesis, of the kind associated with Andy Clark, among others). Perhaps I normally rely upon such information to trigger relevant autobiographical memories. The Korsakoff's suffer, or the participant in a split-brain experiment, may be engaged in entirely normal kinds of explanation-giving. The only difference between them and us is that they are cut off from crucial sources of information, which renders their explanations highly unreliable.

There is another reason to doubt that a unified explanation can be given of the entire range of (what Hirstein calls) confabulation syndromes. Some confabulations are much stranger than others, and much harder to explain. For instance, it is far from clear that Capgras sufferers wholeheartedly believe their delusions, as Hirstein contends (he consistently overestimates the extent to which their behavior is consistent with the content of their delusions). Moreover, some (supposedly) confabulatory syndromes leave the person more or less intact, accentuating the oddness of their claims. Sufferers from Korsakoff's syndrome appear no longer to have the resources to detect the inconsistencies and the implausibilities in their claims, since they cannot remember them for long enough. But some sufferers from anosognosia are globally rational and generally unimpaired. How does the malfunction of a domain-specific checking mechanism result in claims as wild as the claim, on the part of a blind person, that their eyesight is fine? Such a claim has entailments which spread across many domains, and a globally rational person ought to be able to detect the inconsistencies between their claims and their behavior. We have little evidence that our checking processes are as discrete as Hirstein's hypothesis requires.

It may be that some confabulations are explained by entirely normal inferential procedures, whereby we all attempt to construct the narratives of our lives for ourselves and for others, and that others are not explained by any mechanism Hirstein canvases. Some confabulations -- Korsakoff's, and perhaps some of the delusions -- may be the product of relatively normal attempts to explain abnormal experiences, whereas others seem to be produced by abnormal mechanisms, which have yet to be fully comprehended.

Rather than see Hirstein's book as advancing a credible theory of confabulation, I prefer to see it as a collection of stimulating, sometimes mutually exclusive, hypotheses and speculations, some of which may constitute partial explanations for some of the syndromes he examines. There is a great deal of stimulating material in Brain Fiction, and a number of ideas and speculations worth exploring, both in the laboratory and using the tools of philosophy. For all its many virtues, however, it does not present us with a satisfactory theory explaining the entire range of confabulatory syndromes, or even give us much reason to think that there is such a theory to be had.

© 2005 Neil Levy

Dr Neil Levy is a fellow of the Centre for Applied Philosophy and Public Ethics at Charles Sturt University, Australia. He is the author of two monographs and over a dozen articles and book chapters on Continental philosophy, ethics and political philosophy. He is currently writing a book on moral relativism.