How accurate are people at knowing what is happening inside their bodies?

Were people utterly inaccurate at judging their body state and reporting on it, clinical medicine would be deprived of a critical tool.  Evidence has accumulated that in certain circumstances, some people are evidently able to access information about the physiological processes inside of their bodies, and to report on it.  Experiments seem to demonstrate that some people are relatively accurate perceivers of symptoms or physiological state (Jones and Hollandsworth, 1981), (Adam, 1998), and that subjects can be ranked into good or poor estimators of internal state; for instance, with perceivers of heart rate (Schandry, 1981).

When we are actually aware of specific processes inside our bodies and can state this verbally, it would seem that in some fashion unconscious information (or unconscious “information”) has generated or has been transformed into knowledge. However, there is contradictory evidence about accuracy of symptom perception: how good people really are at perceiving various physiological states, and how accurate symptom-reports or other verbal-reports actually are. Many studies have yielded data consistent with the idea that people are not particularly good at accurately reporting on their symptoms or physiological states (Pennebaker, 1982). It is worth pointing out the assertion that people are generally inaccurate about knowing about physiological processes in their bodies reformulates the principle that humans lack epistemological privilege concerning introspective or verbally reported data. In considering the question: are we are likely to be in error when we report on the contents of what is in our minds, or not, it is critical to appreciate the persuasive interpretation of experiments written up in papers such as “Telling More than We Can Know” by the psychologists Robert Nisbett and Timothy Wilson (1977), which seems to show how introspection-based retroactive judgments can are in error.  This category of research typically features subjects placed in circumstances where their choices are influenced by variables controlled by experimenters, and who give explanations for their choosing that display incorrect “folk psychological” constructions. Nisbett and Wilson’s analysis can properly interpreted as to cast doubt on the ability of people to know the causes of our behavior and “higher order” information-processing, and can be summed up with their statement that people may possess “little ability to report accurately on their cognitive processes” (p. 246).

However, I assert that this valuable critique of retrospective judgments has been improperly extrapolated to support a broad skepticism about introspection, what I shall call the “received view” or the “overly skeptical view”, which I might sum-up as the belief that introspective data should generally be regarded with skepticism. As has been noted by careful researchers on introspection, (Schwitzgebel, 2006), this more general rejection of introspection certainly goes beyond what Nisbett and Wilson argued: while they do indeed assert that the evidence of numerous studies shows people are poor at using verbal report-based introspection to the cognitive process behind our judging and deciding, they do not support a general disdain for introspective data. Rather, they state that instead of arguing that introspective reports should simply be discredited, while people do not have introspective access to the cognitive processes, they do have such access to the contents of their cognitions. For instance, Nisbett and Wilson (pg. 255) state that introspection can yield forms of knowledge: knowledge about cognitive content, as an everyday person:

“…knows what his current sensations are and what almost all psychologists and philosophers would assert to be “knowledge” at least quantitatively superior to that of observers concerning his emotions, evaluations, and plans”

Furthermore, the “received view” that introspective reports are to be generally regarded with suspicion is in tension with the clinical use of patient introspection, as well as the high accuracy ratings sometimes displayed in experiments where subjects are asked to evaluate their own physiology. Therefore, while showing appropriate regard for data suggesting limits on introspective access to cognitive information (indeed I will suggest that models of body-knowledge should account for this data), I will nonetheless highlight certain clinical and experimental data that support the following assertion, which  contradicts the view that introspective data should be generally regarded with skepticism:

There exist cognitive processes that allow people to access internal body-state or physiological information in a way that enables fairly, or even highly, accurate verbal reports.  Insofar as this is true, people evidently have some degree of epistemologically privileged access to internal body state or interoceptive information. This relative privilege allows for knowledge of the body, as distinct from mere beliefs.

However, if this is true, some accounting of to what degree or how true it is, with which mitigating conditions, and with what reference to underlying cognitive and neurophysiological mechanisms would be necessary.  For that matter, even if true, demarcating the explanatory power of this principle relative to data adequately explained by the “received view” or “overly skeptical view” is of critical importance. It may be that only special or rare abilities are at issue here, and that the people who have privileged access to their internal physiological information are outliers.

more on the status of introspection in psychology and in neuroscience

An index of the status of introspection within psychology comes from Medin, Markman, and Ross (2004) in the textbook Cognitive Psychology, which notes (pg.20) that:

“Although introspection is not an infallible window to the mind, psychological research is leading to principles that suggest when verbal reports are likely to accurately reflect thinking”

These perspectives all can be said to implicitly or explicitly challenge what I shall call the “received view” or the “overly skeptical view”, which is an interpretation of the Nisbett and Wilson work that goes beyond what those authors’ famous paper actually said. While it is the case that the “Telling More than we can Know” Nisbett and Wilson paper argued persuasively that introspection-based reports of subjects asked to retrospect on the causes of their behavior are generally not accurate, these authors made a point of not dismissing the value of introspection and verbal reporting on the contents of cognition one is aware of , such as sensation or perception and “private facts”. But the “received view” of their research all too often neglects or ignores the more nuanced and balanced view about introspection of the authors, as well as that of other cognitive scientists who carefully investigated the issues involved, such as Anders Ericsson and Herbert Simon (1993).   This is an important concept: see Eric Schwitzgebel’s excellent take on the “Nisbett-Wilson myth“.

What is the most important concept to take away from the controversies about introspection? Probably it is that insofar as researchers want to be able to take advantage of all possible tools and data sources to make sense of the complex, enigmatic processes characterizing body knowledge, they should follow the example set by many physicians and some experimentalists, and be willing to get data by asking subjects or patients for their observations on body state. But here I will go one step further, and assert that the accuracy, or lack of accuracy, of verbal report data relative to other data, can serve as that which must be explained by a comprehensive and robust model of personal or self-reportable knowledge of the body. Doing so would require experiments where verbally reported data might be compared to, and possibly integrated with, data from external sources, such as from brain measurement: “neurophenomenology” in operationalized form.

One such effort came from a trio of researchers interested in assessing whether introspective data on pain had measurable neural correlates (Coghill, McHaffie, Yen, 2003, pg. 8538):

“Using psychophysical ratings to define pain sensitivity and functional magnetic resonance imaging to assess brain activity, we found that highly sensitive individuals exhibited more frequent and more robust pain-induced activation of the primary somatosensory cortex, anterior cingulate cortex, and prefrontal cortex than did insensitive individuals. By identifying objective neural correlates of subjective differences, these findings validate the utility of introspection and subjective reporting as a means of communicating a first-person experience”

This forward-looking research in effect turns behaviorism on its head: instead of verbal reports being rejected or at best tolerated within the overall context of strict objectivity, the very phenomenon the model seeks to explain is “subjective”!

Jack and Roepstorff on introspection

From Trusting the Subject (2003), Anthony Jack and Andreas Roepstorff write:

“The unique challenge facing a science of consciousness is that that the best instrument available for measuring experience depends on cognitive processes internal to the subject. So just how much faith can we place in the capacity of the mind to understand itself? In principle, the construction of a maximally robust methodology for introspective evidence would require a detailed understanding of the operation of introspective processes — the processes that mediate the acquisition of introspective knowledge and underlie the production of introspective reports”

And:

“It is important to realize that no principled problem stands in the way of the scientific assessment of various types of introspective evidence. The testing of the reliability, consistency and validity of various types of introspective report measures lies well within the orbit of currently available methods. A measure may be called ‘reliable’ if it yields the same results when tested in multiple sessions over time (‘test–retest reliability’) and across individuals (a cousin of ‘inter-rater’ and ‘inter-observer’ reliability). Of course, subjects’ reports may differ, and so appear to be unreliable, simply because their internal mental processes and states vary. Thus it is critical to establish well controlled experimental conditions for eliciting reports. The considerable advances in behavioural science since the time of the Introspectionists offers experimenters considerable advantages in this regard (see Ericsson, this volume). Not only do these advances make it much more probable that experimenters can establish conditions under which introspective measures can be shown to be reliable, they also provide much greater insight into the behavioural and neural correlates of experiential phenomena.

A measure may be called ‘consistent’ when it can be shown that the results are not due to specific features of the measurement technique. Tests of consistency provide a means of checking that the observed effect is not due to a methodological artefact. Thus we might test the consistency of introspective  evidence by comparing immediate forced-choice button-press reports with retrospective and open-ended verbal reports. In this way we might establish, for instance: that the results of forced-choice button-press reports have not been influenced by variations in the criterion for response or by automatisation of response such that they no longer constitute true introspective reports; and that retrospective reports have not been distorted by forgetting or memory interference effects.

‘Validity’ is the most important factor to establish, yet it is also the most theoretically complex, and a particularly vexed issue in cognitive science. A measure is validated when it can be shown to accurately reflect the phenomenon it purports to measure. Validity is complex because scientific measures are often simultaneously interpreted as providing evidence for phenomena at a number of
different levels. A rough characterisation of three major sources of evidence in cognitive science might read as follows:

-Data from functional Magnetic Resonance Imaging (fMRI) serves most directly as evidence of cerebral blood flow (which has been validated), less directly as evidence for neural activity (which is in the process of being properly validated), and least directly as a means of identifying and localising specific cognitive functions (far from well validated).

– Behavioural measures (e.g. the averaging of reaction time measures over multiple trials) serve most directly as evidence for stable patterns of behaviour, less directly as a means of assessing information processing, and least directly as means of establishing the existence and operation of specific cognitive functions.

-Introspective reports serve most directly as evidence about the beliefs that subjects have about their own experience, less directly as evidence concerning the existence of experiential phenomena, and least directly as evidence concerning the operation of specific cognitive functions.