The Unreliability of Introspection

Summary

The interview challenges the common assumption that we have especially reliable, direct knowledge of our own conscious experience. Philosopher Eric Schwitzgebel argues that people may be reasonably confident about broad features of experience—such as what they were generally thinking about—but much less dependable when describing finer details. Introspection can be influenced by attention, memory, expectations, language, and ideas about what experience is supposed to be like.

1. The limits of Descartes’ certainty

• Schwitzgebel grants that some thoughts are “self-fulfilling”: having a thought such as “there is a thinker” guarantees that a thinker exists, and uttering “I’m saying these words” guarantees that the utterance is occurring.
• But this does not establish that people can describe their sensory experience with the same certainty.
• There is a significant gap between knowing that one is thinking and knowing exactly what a visual image, pain, or other experience is like.

2. Confidence may depend on the kind of experience

Philosophers often begin with vivid, familiar cases, such as seeing a bright red object or feeling strong pain. These experiences seem difficult to get wrong, especially under ordinary, “canonical” conditions. Schwitzgebel cautions that such examples may not represent introspection generally:

• Mild pain may be confused with an itch or tingle.
• Injured people in high-adrenaline situations may deny pain, though it is not clear that their reports settle whether they felt it.
• People have limited color perception outside the central region of vision, yet may overestimate how clearly they see peripheral colors. In experiments associated with psychologist Edward Titchener, participants kept their eyes fixed while colored cards moved toward the center of their vision. They found it difficult to characterize their peripheral color experience confidently.

Reports about mental imagery may be even less secure. People can disagree about how vivid, detailed, flat, or spatially located their images are, and their reports may not match the images’ actual properties.

3. Perception and judgments about perception

The familiar example of a tilted coin illustrates how difficult it is to say what an object “looks like.” Many people say a tilted circular coin looks elliptical, while also saying that it looks circular in another sense.

Schwitzgebel’s point is primarily about judgments of experience, not necessarily differences in the underlying perceptual experience. What people say may depend on how they think about vision. He suggests that contemporary people may be influenced by analogies to photographs, paintings, and films, where a tilted circle is represented as an ellipse. Other historical traditions used different metaphors, such as ancient Greek comparisons to impressions in wax and early introspective psychology’s interest in stereoscopic vision.

The geometry of a simple “projection onto a plane” account of visual appearance creates difficulties. It can imply distortions for objects away from the central line of sight that people do not typically report experiencing. Alternative projections, such as onto a sphere, bring their own problems. These examples make claims about apparent shape, depth, and flatness contestable rather than self-evident.

4. Memory and the study of experience

Reports about an experience can be distorted by memory. Luke raises the example of people saying that time slowed during frightening or adrenaline-filled events. One proposed explanation is that the event is remembered as longer because more details were stored, rather than because time literally seemed slower while it occurred.

Schwitzgebel considers that explanation plausible but not established. A single experiment is rarely decisive because researchers must interpret self-reports and make assumptions about how experience relates to behavior and memory. In consciousness research, competing explanations can often be used to defend the accuracy of a report. Schwitzgebel therefore emphasizes the importance of converging evidence from a range of methods rather than relying on any one experiment.

5. The “sparse” and “abundant” views of experience

Schwitzgebel describes two contrasting views about what is present in consciousness when attention is elsewhere:

• Sparse or thin view: Experience contains little beyond what someone is attending to. For example, a person absorbed in driving may not be experiencing the radio or the pressure of the seat in any meaningful sense.
• Abundant or rich view: Experience includes many unattended details, such as the feeling of shoes, background sounds, or other sensory input.

People asked about this issue appear to divide roughly evenly, and often find their preferred answer obvious. Schwitzgebel uses the “refrigerator-light error” to explain a possible source of confusion: opening a refrigerator reveals that its light is on, but that does not show that it stayed on while the door was closed. Similarly, reflecting on whether one is experiencing one’s feet or background sounds may itself bring those experiences into awareness, creating the mistaken impression that they were continuously present.

6. A proposed method—and its limitations

Schwitzgebel describes a way to investigate unattended experience using immediate retrospective reports:

1. Give participants beepers and let them go about their ordinary activities.
2. Have the beeper sound at relatively long, unpredictable intervals.
3. Ask participants to recall the last undisturbed moment immediately before the beep.
4. Ask targeted questions about that moment—for example, whether they had a tactile experience in a particular foot.
5. Use different prompts for different participants, so the study probes specific aspects of experience.

This approach aims to avoid the problem that asking people to inspect their experience in the moment may create the very experience being investigated. However, it remains imperfect:

• Participants may confuse what was happening in the environment with what they experienced.
• They may create or elaborate an image while remembering it.
• They may forget an experience that was present only briefly.
• Brain measurements and ordinary performance tests do not resolve the issue without an account of how brain activity, behavior, memory, and conscious experience relate. Using those measures may presuppose the very answer—sparse or abundant experience—that the study is trying to establish.

Schwitzgebel says he has used beeper-based reports but interprets the results cautiously. He sees no clearly superior short-term method, while remaining open to the possibility that future science could make progress.

7. The study with Russell Hurlburt and “Melanie”

Schwitzgebel and psychologist Russell Hurlburt collaborated on Describing Inner Experience: Proponent Meets Skeptic.

• Hurlburt favored a beeper-based method: participants report on the last moment before a beep, and an interviewer probes their account.
• Schwitzgebel was skeptical of how accurately people could describe their experience, including through this method.
• They interviewed a participant called Melanie, presenting edited transcripts of her reports and their discussions about them.
• They agreed that reports may have some initial credibility at a broad level—for example, that a person was thinking about a novel or had an image of fighter planes—but disagreed about how far that credibility extends.
• Schwitzgebel thinks reliability declines quickly as questions become more detailed or theoretically significant, such as asking how vivid an image was or whether a person also had visual sensory experience of a page.

8. Implications for epistemology

Schwitzgebel believes these doubts challenge traditional foundationalism, which treats knowledge of immediate experience as a secure basis for knowledge of the external world.

He does not conclude that we should be skeptical about the external world. Instead, he suggests that knowledge of ordinary objects may be more dependable than knowledge of the precise character of our experience.

For example, someone may know that a tissue box is on a table and use that knowledge, together with assumptions about how vision works, to describe their experience of the box. Their knowledge of the object need not depend on first having an accurate introspective account of their experience.

A causal or reliabilist account of knowledge may allow perception to produce reliable judgments about the world without requiring a separate, accurate appreciation of the phenomenal qualities of the experience itself.

Speakers and sources featured

Speakers in the video

• Eric Schwitzgebel — philosopher interviewed in the video.
• Luke Muehlhauser — interviewer.
• Unidentified narrator — introduces Schwitzgebel at the beginning.

People, works, and sources discussed or referenced

• René Descartes — the cogito and the claim that knowledge of thought provides a foundation for knowledge.
• Edward B. Titchener — introspective psychology and demonstrations involving peripheral color vision.
• John Locke and David Hume — views about visual experience and whether objects appear flat.
• Sextus Empiricus and Epicurus — ancient Greek discussions of perception and visual appearance.
• Michael Tye and Alva Noë — philosophers associated in the interview with contemporary views about tilted objects appearing elliptical.
• Russell Hurlburt — co-author with Schwitzgebel of Describing Inner Experience: Proponent Meets Skeptic.
• Melanie — the participant whose reports are examined in that book.
• RadioLab — a podcast episode cited in the discussion of time perception during frightening experiences.
• Change-blindness research — referenced as evidence that people may fail to notice or remember details in their surroundings.