Confabulatory explanations

Here are some examples of confabulatory explanations in the clinical population. In anosognosia people deny some serious impairment. When a person with a paralysed leg is asked why she cannot climb stairs, she may say she suffers from arthritis and she is less mobile as a result. In the Capgras delusion, people believe someone closed to them has been replaced by an impostor and may suggest that there are minor physical differences between the loved one and the alleged impostor (“his eyes are too close”).

We are all vulnerable to confabulation when we are asked why have a certain attitude, such as a moral judgement, a preference, or an emotional reaction. We make up a reason for it that seems plausible but does not track the causal mechanisms responsible for the formation of that attitude. My favourite example comes from a classic study by Nisbett and Wilson (1977), where people choose between identical pairs of socks (they prefer items on their right), and then are asked the reason for their choice. They answer that the chosen item is softer or has a brighter colour.

Non-clinical confabulation can occur when wildly implausible belief states are induced to model delusions in the lab. Somatoparaphrenia is the delusional belief that one’s limb belongs to someone else and can be modelled with hypnosis (Rahmanovic, Barnier, & Cox 2010 and 2012). Highly hypnotisable participants receive a suggestion that their non-dominant arm belongs to someone else and an instruction to forget the hypnotist giving them that particular suggestion. They are then asked to pick up objects on a tray located next to the ‘foreign’ arm. If they use the other arm, they are asked why. They say that the arm closer to the tray is tired or feels heavy. Similar confabulations are found in people who report hypnotic analogues of the delusion of mirrored-self misidentification (see this paper for details). Highly hypnotisable participants deny that the image in the mirror is an image of themselves, and defend their belief against challenges by saying that the alleged stranger in the mirror is older than they are or has eyes of a different colour from theirs.

Confabulators fill explanatory gaps with inaccurate information, when accurate information is not available to them. Crucially, different senses of ‘available’ apply to different forms of confabulation. Limitations of introspection, memory or self-interpretation are responsible for lack of availability in non-clinical cases, whereas deficits due to brain damage may compromise availability in clinical cases.

It seems to me that, if the ‘true’ explanation of the phenomenon we are asked about is somehow unavailable to us, providing a ‘false’ explanation for it is less epistemically blameworthy. In most circumstances, we should be aware that we are saying something false (the chosen socks are after all identical to the discarded ones, and our eyes are the same colour as those of our image in the mirror), but our ignorance of the true explanation for our attitude is ‘faultless’. We cannot have introspective access to position effects on decision making (we learn about them if we study psychology) or to a hypnotic suggestion we were asked to forget (we learn about it during debriefing).

Providing a false explanation driven by plausibility considerations when the true explanation is not available may have some pragmatic benefits. Offering a confident answer as opposed to saying “I don’t know” has advantages in a social context, contributing to external perceptions of the self and to a positive self-image as coherent believers and decision makers. Especially in clinical cases, the offered explanation may have a defensive function (Ramachandran 1996), preventing confabulators from acknowledging an undesirable truth (e.g., that there is something wrong with them if they cannot recognise their loved ones, or that they are now seriously impaired due to the paralysis of a limb).

Do confabulatory explanations have also epistemic benefits?


  1. Thanks very much Lisa for this , and for all your posts this month. I’ve also enjoyed getting familiar with your work, and looking up some of your references. (It’s helped me significantly while putting together a book proposal.)

    It seems the answer to your closing question is pretty clearly ‘yes’. For instance, one can imagine confabulations supporting a healthy self-image, which then encourages investigating new and controversial ideas. It’s a special case of a useful fiction aiding action (even though the action in this case is an act of inquiry).

    I was also intrigued by your statement “if the ‘true’ explanation of the phenomenon we are asked about is somehow unavailable to us, providing a ‘false’ explanation for it is less epistemically blameworthy.” I’m inclined to agree, on the basis of a principle we might call “epistemic ought implies epistemic can.” (This is a principle I find useful in other contexts as well.) But even if the principle is true, the trick is to tell when it is rightfully applies.

    In some cases, the true explanation of behavior may well be unavailable. Nonetheless, epistemic agents also seem capable of “negligence,” in their effort to avoid self-deception. (I am reminded of W.K. Clifford’s ship owner.) To my mind, then, the real question is how to distinguish cases of negligence from excusable cases. Your different senses of ‘available’ would help I think–though unfortunately, there would probably remain lots of cases that are hard to adjudicate.

    • Thank you Ted!

      I’m tempted to answer yes to the question whether a false self is preferable to a fading self, but I can imagine arguments to the effect that the advantage is more plausible if described in terms of good functioning or subjective wellbeing than in epistemic terms.

      If the sense of self acquired through confabulation lacks ‘correspondence’ with reality to a very large extent, then I guess leaving some key questions about the self unanswered may not be the worst option.

      That said, I’m totally with you re: epistemic version of the ought-implies-can principle. Rochelle Cox and I appeal to it in our Faultless Ignorance paper when we comment on Hirstein’s definition of confabulation.

      I also agree that the difficulty will be in establishing which grounds for explanation are unavailable and what sense of availability applies to each context.

      There is an interesting discussion about whether people with anosognosia realise at some level the nature of the impairment they deny (and thus are to some extent blameworthy for providing alternative explanations for their impaired functioning) in Neil Levy’s paper on self-deception. It’s in the excellent volume on Affective and Motivational Influences on Belief Formation edited by Bayne & Fernandez.

      Similar questions about ‘covert recognition’ have been asked with respect to delusions of misidentification, when people with Capgras delusion are friendly towards the alleged impostor and people who claim that there is a stranger in the mirror (mirrored-self misidentification) give the stranger their own name. Rochelle Cox, Amanda Barnier and I briefly address this in the paper on whether we can recreate delusions in the lab.

Comments are closed.