Introspection and Its Limits
On this page(6)
Someone asks what you are feeling. You answer, and you do not answer the way you would answer a question about what they are feeling. You do not watch yourself, note how you have been behaving, and infer. Something more direct happens, and it happens fast enough that most people never notice it happening at all.
The previous four articles have all relied on this. Every claim about what an experience is like, every report about how the coin looked, came from someone consulting their own case. This article examines the instrument.
The capacity
Definition
Introspection
The process by which a subject comes to know about their own current mental states, apparently without observing their own behaviour and without inference from anything else.
The definition is framed in terms of the appearance rather than the reality, because what introspection actually is remains disputed. It has been modelled on perception, as an inner sense that scans one's own states, and that model has been attacked on the grounds that no inner organ has been found and that the analogy generates problems it cannot solve. It has been modelled on a capacity to redeploy ordinary reasoning on oneself, which has trouble explaining the immediacy. The intermediate tier works through the accounts. What is not in dispute is that people answer questions about their own minds in a way that differs strikingly from how they answer questions about anyone else's.
Three claims that get run together
The traditional position bundles several distinct theses, and separating them is most of the work, because the evidence bears on them very differently.
Definition
Privileged access
The thesis that a subject has a way of knowing their own mental states that no one else has and that does not rely on observation of behaviour.
Definition
Incorrigibility
The thesis that a subject cannot be mistaken about their own current mental states, so that a sincere first-person report about them cannot be false.
These are in descending order of strength and they come apart cleanly. You can hold that a person has a special route to their own mind while allowing that the route sometimes fails, which is privileged access without incorrigibility. You can hold that we defer to people's reports as a social and linguistic practice while denying that anything special is going on epistemically. Incorrigibility entails the other two, and neither of the others entails it.
Descartes took something close to the strongest thesis for granted. The whole strategy of the Meditations depends on the thinker's access to their own thinking surviving doubts that dispose of everything else, and it is that access, not the reliability of the senses, that provides the fixed point.
The case for at least the weaker claims is not a philosopher's invention, and it should be granted before the objections start. You do not work out that you are in pain. A person who said they had inferred their own toothache from their behaviour would have said something bizarre, and if someone else disagrees with you about whether you are in pain, the presumption is heavily on your side in a way it is not for any other subject matter. Something asymmetric is going on and any account has to make room for it.
Check your understanding
What the psychology found
Definition
Confabulation
The production of a sincere, confident report that is not based on the process it purports to report, and which the subject does not recognise as invented.
The word matters because the phenomenon is not lying. A confabulating subject believes what they are saying, offers it without hesitation, and is often puzzled by the suggestion that it might be wrong.
Richard Nisbett and Timothy Wilson gathered the evidence for it in 1977 in a paper that has shaped the debate ever since. In one of the studies they describe, shoppers were shown several identical pairs of stockings laid out in a row and asked which was of the best quality. Choices were heavily affected by position, with the rightmost pair chosen far more often than the leftmost. Asked why they had chosen as they did, subjects gave confident answers about texture and weave. Asked directly whether position might have mattered, they denied it, and some treated the question as odd.
Their general conclusion was that when people report on the causes of their own judgements and choices, they are not consulting the process. They are producing a plausible explanation from their background theory of what tends to influence people, and that explanation happens to be right when the theory is right and wrong when it is not.
The effect has been extended in ways that are harder to explain away. Petter Johansson and colleagues reported in 2005 that subjects who chose between two photographs on the basis of attractiveness could be handed the photograph they had rejected, by means of a sleight of hand, and would usually fail to notice the switch. Many then explained at length why they had preferred the face they had in fact rejected, citing features of it as their reasons.
What that does and does not establish
The results are often taken to show that introspection is unreliable across the board. Read carefully, most of the classic evidence supports something narrower, and the distinction is worth holding on to.
What the subjects were being asked for was a causal explanation of their own behaviour: why they chose this one. That is a question about a process, and the processes that produce preferences are not the sort of thing anyone claimed to have privileged access to. Nobody's traditional view held that you can inspect the mechanism that generated your preference. The claim was about states, not about the machinery that produces them.
So a defender of privileged access has a reply available. Ask a subject in the stocking study whether they preferred the rightmost pair and they will be right. Ask them why and they will confabulate. Preference is available to introspection; its causes are not, and mistaking a report of the second for a report of the first is the error.
That reply is serious and it is not the end of it, because a further line of argument targets exactly the states the traditional view was confident about. Eric Schwitzgebel has argued in detail that people are unreliable even about their current conscious experience under favourable conditions. Ask someone whether their visual experience is detailed and coloured at the edges of the visual field, or how much imagery accompanies their thinking, or what precisely they are feeling in a mixed emotional state, and you get confident answers that conflict between people, conflict with the same person's answers on other occasions, and have historically shifted with whatever theory was fashionable. If people cannot agree about whether visual imagery is like seeing a picture, the confidence with which each of them reports their own case is not tracking accuracy.
Check your understanding
Where that leaves the instrument
The honest summary is graded rather than global, and the grades matter for how the rest of the subject is done.
Access is strong for the fact that one is in some broad kind of state. That you are in pain rather than elated, that you are seeing something rather than remembering it, that you want the thing you are reaching for: reports of this sort are rarely challenged and there is little evidence that they should be.
Access is weak for the fine detail of experience. How much of the visual field is in focus, what a mixed emotion consists of, whether thought is accompanied by inner speech: here the reports vary in ways that cannot all be correct.
Access is very weak for causes and processes. Why you chose, what influenced you, what you would have done otherwise: on these the reports are theory-driven and the theory is often wrong.
That grading is itself a philosophical result rather than a concession. It tells you which introspective claims a theory of mind may lean on and which it may not, and a surprising number of arguments in the field turn out to lean on claims from the second and third categories.
Objection
The concession is fatal and cannot be limited. Everything the philosophy of consciousness works with comes from introspection. The claim that there is something it is like to see red, the claim that the coin looks elliptical, the claim that a physical duplicate could lack experience: every one of these is a first-person report, and there is no other source. If those reports are unreliable in the ways described, the subject has no data. Worse, the unreliability cannot be corrected, because the only instrument available for checking a first-person report is another first-person report.
Reply
The objection is pressed by serious people and there is a version of it nobody has fully answered. Two things limit it.
The first is that the results are not uniform, and the arguments the objection lists depend on the reports that survive. The claim that there is something it is like to see red is a claim of the first kind above, not a detailed judgement about the structure of the visual field. Schwitzgebel's own conclusion is that we are unreliable about the character of experience while its existence is not in question, and a critique that assumed otherwise would be a different argument requiring different evidence.
The second is that the checking problem is not unique and is not insoluble. Instruments that cannot be checked against a direct reading of what they measure are checked in other ways: against each other, against theory, and by finding the conditions under which they go wrong. That is exactly what the confabulation research does. It uses behavioural evidence to identify circumstances in which reports come apart from what they report, which is a way of calibrating introspection from outside without ever stepping outside the mind.
Where the objection genuinely bites is against a certain style of argument. Elaborate conclusions have been drawn from finely detailed claims about how experience is, offered as though they were obvious on inspection and not open to reasonable disagreement. When two careful philosophers report their own experience differently and neither can be shown to be wrong, that is a sign the method has been pushed past what it can support, and it happens more often than the field admits.
Check your understanding
Next
Whatever its limits, introspection gives each person a source of information about one mind that is unlike anything they have about any other. Everyone else is available only through behaviour, and the question of what entitles anyone to think there is anything behind that behaviour is the next article.
Further reading
- Richard E. Nisbett and Timothy D. Wilson, "Telling More Than We Can Know: Verbal Reports on Mental Processes", Psychological Review, volume 84, number 3 (1977), pages 231 to 259.
- Petter Johansson, Lars Hall, Sverker Sikström, and Andreas Olsson, "Failure to Detect Mismatches Between Intention and Outcome in a Simple Decision Task", Science, volume 310 (2005), pages 116 to 119.
- Eric Schwitzgebel, "The Unreliability of Naive Introspection", The Philosophical Review, volume 117, number 2 (2008), pages 245 to 273.
- Stanford Encyclopedia of Philosophy, "Introspection".
- Stanford Encyclopedia of Philosophy, "Self-Knowledge".
Quiz yourself on this article
9 questions. Miss one and it comes back until you get it right.
Reached the end?
Mark it, and the sidebar keeps track of where you have been.