Explaining the placebo effect: Aliefs, beliefs, and conditioning

Abstract

There are a number of competing psychological accounts of the placebo effect, and much of the recent debate centers on the relative importance of classical conditioning and conscious beliefs. In this paper, I discuss apparent problems with these accounts and with “disjunctive” accounts that deny that placebo effects can be given a unified psychological explanation. The fact that some placebo effects seem to be mediated by cognitive states with content that is consciously inaccessible and inferentially isolated from a subject's beliefs motivates an account of the placebo effect in terms of subdoxastic cognitive states. I propose that aliefs, subdoxastic cognitive states that are associative, automatic, and arational, can provide a unified psychological account of the placebo effect. This account also has the potential to illuminate interesting connections to other psychological phenomena.

Keywords:

• Alief
• Belief
• Conditioning
• Placebo Effect

Acknowledgements

I would like to thank Tamar Szabó Gendler and Esther Sternberg whose talks at “The Study of the Human Self” conference at William & Mary in fall 2008 prompted me to try to explain the placebo effect in terms of aliefs. Thanks also to Paul S. Davies for helpful feedback on an early draft and to Stephen Crowley for comments on a shorter version of this paper, which I presented at the 2010 Society for Philosophy and Psychology meeting in Portland, OR. I am grateful for financial support from a Faculty Summer Research Grant from the College of William & Mary.

Notes

Notes

[1] Some authors (e.g., Hoffman et al., 2005) have regimented terminology as follows. The “placebo response” occurs in a given individual and is the difference between the untreated, natural history condition and the individual's condition after administration of a placebo (i.e., after interaction with elements in the psychosocial context). By contrast, the term “placebo effect” refers to the average response of members of a group to placebo manipulation (i.e., the average placebo response for a particular group of subjects). Not all authors adopt this usage, and I occasionally depart from it in this paper.

[2] Gendler does not claim to have discovered a new category of mental states. Rather, she claims merely “to have noticed a certain commonality across some lines of thought that might otherwise have appeared disparate” (2008b, p. 557). In this paper, I argue that this commonality extends to the placebo effect, as well.

[3] There is compelling brain imaging work that supports the view that the placebo effect involves an actual reduction in the intensity of experienced pain and not merely in a reduction in the amount of pain that is reported (e.g., Wager et al., 2004).

[4] As has been repeatedly pointed out in the placebo literature (e.g., Hoffman et al., 2005; Stewart-Williams and Podd, 2004; Voudouris, Peck, & Coleman, 1990), the conditioning and expectancy accounts are not mutually exclusive, since response expectancies can be developed through conditioning. However, even Kirsch and colleagues admit that some conditioning phenomena—including conditioned immunosuppression, conditioned taste aversion, and subliminal conditioned stimuli—do not appear to be cognitively mediated (at least by conscious beliefs or expectancies; more on this below).

[5] An anonymous referee noted that someone could hold that conditioning is mediated by representational states other than beliefs or desires, and the resulting view would not collapse into a cognitivist view like those discussed in section 1.2. I agree, and I think that aliefs should be preferred to beliefs in any such account, for the reasons given in sections 3.2 and 3.3. See also section 5.1.

[6] “Placebo responses are mediated by conditioning when unconscious physiological functions such as hormonal secretion are involved, whereas they are mediated by expectation when conscious physiological processes such as pain and motor performance come into play, even though a conditioning procedure is performed” (Benedetti et al., 2003, p. 4315).

[7] Perhaps there are some unconscious beliefs that cannot be brought to awareness, or at least not readily, e.g., those that are repressed. However, in these cases, there is plausibly some mechanism that intervenes to block conscious access (Stich, 1978, p. 505).

[8] If this latter description is correct, as I argue below, then exposure to the formal features of a treatment unconsciously “primes” subjects in much the same way that performing scrambled sentence tasks does. Such priming results in automatic, unconscious activation of the associated affective and behavioral routines. See, e.g., Bargh, Chen, and Burrows (1996), and Gendler (2008a, pp. 656–661).

[9] Following Stich, I assume that inference is not by definition a relation solely between beliefs (1978, pp. 507 & 511–517).

[10] Note that Gendler uses the term “content” in a somewhat idiosyncratic, general way that may include affective states and behavioral dispositions (2008a, p. 635). As I discuss below, the contents of aliefs involved in the placebo response also involve physiological dispositions that one may be reluctant to call behavioral. Also, even if an alief is consciously activated, usually not all of its content is consciously accessible.

[11] One may also want to allow a role for the subdoxastic analog of desire, which Gendler dubs “cesire” (2008a, p. 642), for similar reasons to those given by Price and Fields (1997), mentioned above.

[12] These descriptions of the representational, affective, and physiological components of the alief's content are stilted and artificial, and for good reason: I suspect that the contents of the relevant aliefs are non-propositional and resist description in natural language.

[13] In fact, there is some intriguing, although almost anecdotal, evidence that suggests that one can experience a placebo response even if one is convinced (Park & Covi, 1965), or has some reason to believe (Bergmann et al., 1994), that one is receiving merely a placebo treatment. If such cases are corroborated, then they provide another example of alieving a content (the pill will make me better) without accepting it, and thus not believing it (see Gendler, 2008a, pp. 648–651).

[14] However, the aliefs involved in some placebo responses do have behavioral content. For example, in a double-blind study, McRae et al. (2004) performed sham surgery for some patients with Parkinson's (i.e., brain surgery was performed on all subjects but the half in the placebo arm did not actually receive transplants of embryonic dopamine neurons). Even the subjects who received the placebo surgery showed significant improvement in physical, emotional, and social functioning, even after one year.