Schachter and Singer’s famous adrenaline study aimed to collect evidence in favor of a two-factor theory of emotion, according to which a changed state of arousal leads agents to form feelings with a given valence that depends only on the epistemic/motivational context. Participants’ arousal was manipulated by injecting them, under pretext, with adrenaline or a placebo. Only a subgroup of the adrenaline participants were informed that they had received a drug that would modify their arousal level. Participants were subsequently invited to stay in a waiting room where a confederate was either pretending to be euphoric or angry. Participants’ emotional responses, observed in their behavior and subsequent self-report, differed in the various conditions: those unaware of having been injected with adrenaline, and placed in the anger condition, felt angriest, followed by the placebo + anger subjects. The least angry were the adrenaline informed participants. In the euphoria condition, misinformed adrenaline participants were “somewhat” happier, adrenaline informed ones somewhat less happy (in the euphoria condition, the results failed to reach significance both for behavior and self-report).
Were Schachter and Singer successful in making the point that valence of a feeling is a matter of attribution of the source of an experienced arousal? Several powerful objections have been raised against this claim. Recall that subjects were asked to what degree they would describe themselves as happy or angry. A first problem is that the questionnaire suggested the relevant target categories of emotions, which is disturbingly close to influencing participants’ responses (see Plutchik & Ax 1967 and Gordon 1987, p. 100). Furthermore, as noted above, ex post-facto reflective labeling of one’s emotion does not need to express one’s original feelings. As shown by Nisbett & Wilson (1977), self-reporting is highly sensitive to rationalizations from context. A second problem, mentioned by the authors in the discussion, is that the subjects’ verbal reports and emotional behavior failed to confirm expectations in the euphoric condition. A third methodological problem, also recognized by the authors, is that the student participants had their own independent reasons for feeling anger in passing this longish test, which predisposed them to feel anger. There are, however, more theoretical objections.
On Schachter and Singer’s view, the core feeling of an emotion is an arousal change, which can be artificially induced by drugs. Valence is supposedly gained through contextual beliefs and motives. If this view is accepted, why should we expect that contextually relevant beliefs specify the feeling itself (e.g., the anger experience)? Participants may indeed have been led to believe that they were angry when they were actually merely aroused. This does not show, however, that they ever felt anything else than an arousal change (Gordon 1987, pp. 100–101). Schachter and Singer may have only biased self-attributions and self-report toward target emotions. The behavioral changes that were observed and attributed to felt emotion, in addition, can be imputed to social influence, rather than to intrinsic changes.
A final worry is that inducing in a participant a somatic marker normally associated with a given feeling (e. g., increased heart rate), and providing the person with a context rationalizing this somatic change, does not amount to an ecological way of producing a feeling. A cognitivist theorist of emotion will insist that the mere association between a physiological cue of the feeling f and a context does not amount to the realization, by a participant, that she feels f because she is in such and such a context (Gordon 1987, pp. 98–99).[25] As discussed in section 3, the expressive mode has a nonconceptual representation of this causal connection. The architectural relation between feelings and affordances explains why subjects experience a systematic connection between their feeling and what it is “about”, much in the same way that an agent experiences a systematic connection between an intention to move and the goal that is aimed at—that is, without needing to represent conceptually the causal connection between the two. Nothing prevents the emoter, however, from forming a secondary conceptual representation of the emotional experience she has had, and reappraising the context on the basis of her background beliefs. As a consequence of this concept-based reapparaisal, the agent may either discount the relevance of her initial feeling (as in the fear-of-snake case), or redescribe it in the richer terms that she now has available (as was done, presumably, by the Schachter and Singer participants).
Taken together, these objections have led most theorists to reject Schachter and Singer’s two-factor theory of emotion, and to look for alternative accounts of the role of inferences in self-attribution of feelings. It is interesting to see, however, that a two-factor theory has also been applied to the case of M-feelings.