Guts, Nerves, and Zeal:
A New Adrenaline Objection to Prinz on Emotion
Mikio Akagi
University of Pittsburgh
Abstract. Jesse Prinz holds that emotions are perception-like appraisals that are
identical to detections of physiological changes in the body. Prinz defends his
view against the adrenaline objection, based on the famous 1962 study by
Stanley Schachter and Jerome Singer. However, Prinz misunderstands the
structure of the study. I argue that Schachter and Singer present evidence that
there can be detections of physiological changes in which emotional responses are
suppressed by cognitive states such as beliefs. If there can be detections without
emotions then, contra Prinz, emotions cannot be identical to such detections.
Prinz underestimates the extent of interaction between emotion and cognition.
1. Introduction. Jesse Prinz argues that emotions are perception-like appraisals
of organism-environment relations that track bodily changes. By identifying
representational appraisals with detections of physiological states, Prinz tries to
achieve a synthesis of cognitive appraisal theories and somatic theories of
emotion: an embodied appraisal theory. However,
weak against an alternative that accommodates more interaction between
view allows that cognitive states can elicit emotions, it does not allow that
cognitive states can suppress emotional responses when the eliciting conditions
are otherwise sufficient. However, there is empirical evidence that emotional
responses can in fact be suppressed through the manipulation of cognitive states.
In the
embodied appraisal
account of emotions. In Section 3 I will describe what Prinz calls the adrenaline
objection
classic study
characterization of the study is flawed. A more faithful understanding of the
according to which there can
be physiological state-detections without emotions. This would be a
Section 4 I will discuss a series of potential replies to the new adrenaline
ion. I
between cognition and emotion.
2. Emotions as Embodied Appraisals. Prinz presents his account of emotion
as a grand reconciliation between somatic theories as defended by Antonio
Damasio (1994) and Robert Zajonc (1984), and cognitive appraisal theories as
articulated by Richard Lazarus (1984). Roughly, somatic theories take emotions
Akagi 2
to be detections by neural systems of changes in the body changes in heart rate,
hormone levels, facial expressions, &c. Emotions are not just any detection of
this is registered by a neural system, it is not therefore an emotion. The
particular physiological configurations that trigger emotional responses must be
discovered empirically.1 Cognitive theories, on the other hand, take many shapes,
paradigmatically holding that emotions are forms of judgment (Nussbaum 2001).
Prinz is particularly interested in what he calls cognitive appraisal theories,
where an appraisal is a representation of the relation between an organism and
an environment that bears on well-being (Prinz 2004, 51). Lazarus claims that
emotions are appraisals that represent core relational themes, such as danger,
real o
Distinct themes
correspond to distinct emotions in these cases fear, sadness and disgust,
respectively.
Unlike Lazarus, however, Prinz does not take emotional appraisals to be
cognitive states, i.e. states involving cognitive representations (Prinz 2004, 64
66).2
ognitive representations are those representations that an
agent can activate directly, as by what might be called an act of the will.
Linguistic concepts are paradigmatic cognitive representations. Emotions are
clearly not cognitive representations in this sense, since whether or not we feel an
emotion is not directly subject to the exercise of the will. Emotions would be
cognitive states, however, if they were constituted in part by cognitive
representations, such as concepts. Prinz argues that while some emotional
episodes are clearly accompanied by conceptual representations, emotions are not
essentially cognitive. He cites evidence reported by Zajonc (1984) and others that
emotional states can be induced without any cognitive mediation.3
his 2002 for an extended treatment), and in particular does not seem to
correspond to the usage of the term in cognitive science. However, I will adopt
Prinz arranges his reconciliation by suggesting that
appraisals on the model of perception. On this view, the
constitute emotions are those that distally represent
themes by proximally tracking physiological state changes in the
Basic emotion:
(
)
1
we understand
appraisals that
core relational
body.
/emotion
Thanks to Brian Ballard for promoting clarity on this point.
2
(Prinz 2004, 52 55).
3
A flagship example is what might be called short-circuit ophidiophobia, a fear
response triggered (normally by visual experience of snake-like objects) in early
visual processing and mediated by the amygdala independently of cortical areas.
Akagi 3
That is, physiological changes bear reliable relations to core relational themes,
and emotions are mental states that reflect those relational themes through
sensitivity to physiological changes. So for example, the sight of a snake slithering
nearby causes inter alia an increase in heart rate, that are detected and which
detection constitutes a fear response (Prinz 2004, 69). The essential move in his
reconciliation is to identify emotions both with appraisals of core relational
themes (henceforth appraisals ) and with detections of bodily changes
(henceforth
). The result is an embodied appraisal account of emotion
(Prinz 2004, 68 69, 77 78). I will argue that the central identification of
emotions with detections is poorly motivated.
Before moving on, though, I should address one significant complication to
this picture that renders it much more plausible. Obviously, we respond
emotionally to stimuli that are not obviously perceptual. For example, a fear
response might be triggered by thinking about an exam or a looming deadline.
Prinz claims that cognitive states like believing that a deadline is approaching
can come to function as eliciting conditions for emotions through learning. In
effect, Prinz claims that cognitive states can come to replace perceptual states as
eliciting conditions for physiological changes:
Derived emotion:
parasitic on cases in which emotions
77, 68). We feel fear when we imagine the deadline
because deadline-thoughts have tapped into the older control mechanisms that
responded to things like snakes.
3. The Adrenaline Objection. If, as Prinz has it,
78), then the physiological concomitants
of emotion must be rich and varied enough that categorial differentiation between
them is sufficient for individuating distinct emotions. For example, the
physiological correlates of fear and anger must be different enough that a fear
response but not an anger response is reliably set off by one complex of
physiological responses, and vice versa for a different complex of physiological
responses. The empirical evidence on this matter is not decisive, although Prinz
provides some reasons to think that he will be vindicated (2004, 72 74). The
most significant empirical challenge to the existence of emotion-specific
physiology is the adrenaline objection, which is based on evidence from several
studies culminating in the famous study by Stanley Schachter and Jerome Singer
(1962).
Schachter and Singer attempted to elicit different emotional responses
(euphoria and anger) by manipulating cognitive judgments while keeping
physiological states constant. In their study, they injected participants with
epinephrine (adrenaline), either informing participants about the physiological
effects of the injection or omitting these details. These conditions are referred to
Akagi 4
by Schachter and Singer as the epinephrine-informed and epinephrine-ignorant
conditions. There was also a placebo condition in which participants were
injected with an inert saline solution. They then placed participants in a room
with distractions and an amusing confederate, or an insulting questionnaire and
an infuriated confederate. The upshot is supposed to be that given the same
physiological stimulation (epinephrine), participants could interpret their own
physiological arousal as either euphoria or anger depending on the social context:
Different contexts cause the same physiological state (e.g. heart
palpitations) to be interpreted in different ways. It is the difference
in interpretation, not physiology, that accounts for the difference in
emotional state. (Prinz 2004, 70)
Prinz replies to this objection with three worries:
First, (P1) [Schachter and Singer] assume that the subjects who
display distinct emotional behavior are actually in different
emotional states. Second, (P2) they assume that the subjects who
display different emotional behavior are actually in the same
physiological state. Both assumptions can be challenged. (2004, 70;
numbering mine)
These objections will be discussed in more detail in the next section.
worry is that (P3) even if both these assumptions are granted, it is not fatal to
his account that appraisal/detections may sometimes occur in the absence of the
bodily chan
In short, there can be false
positives, especially when hormones are involved. This concession is not fatal to
identification of emotions with detections as long as there are false
detections in the relevant cases.
However,
theory from the design of their famous study. Prin
reflects the former, but is not faithful to the latter. I claim, following Schachter
and Singer themselves, that the main manipulation in the study is not between a
euphoria condition and an anger condition in one experiment, but between the
informed and uninformed conditions in two experiments.
interpretation, the main manipulation of the study is between the anger and
euphoria conditions for epinephrine-ignorant participants, whereas participants
in the epinephrine-informed and placebo conditions should be considered control
groups for comparison. But this interpretation makes a mess of Schachter and
-informed
group. Instead, we should think of the participants in the epinephrine-ignorant
condition as a comparison group along with the placebo group. The main
manipulation is between the epinephrine-informed and the epinephrine-ignorant
groups. It is between these conditions that analytical comparisons are made by
Schachter and Singer and that they report significance figures. They never report
The result of
the study, understood aright, is not that identical physiological states elicit
distinct emotions, but that in the epinephrine-informed condition emotional
responses are suppressed in conditions that, had the participants not been
Akagi 5
informed about the side-effects of epinephrine, would have been sufficient to elicit
emotional responses. Even participants in the placebo condition exhibited greater
emotional responses than the participants in the epinephrine-informed condition.
This experimental design was successfully used in a euphoria-eliciting scenario,
then repeated for an anger-eliciting scenario.
just criticisms of the theory of emotions
endorsed by Schachter and Singer. They do assume that because they had no
physiological measures to distinguish participants in the euphoria and anger
conditions, those participants all share the same physiological state. Furthermore,
Schachter and Singer do take their study to be evidence that given a state of
physiological arousal for which an individual has no explanation, he will label this
state in terms of t
In other words, that emotional states can be artificially elicited when subjects
misidentify the cause of their arousal. This conclusion is too strong; it is
and has been challenged by subsequent
studies (see Reisenzein 1983 for a review). However, Schachter and Singer also
find that
given a state of physiological arousal for which the individual has a
completely satisfactory explanation, he will not label this state in
terms of the alternative cognitions available. (Schachter & Singer
1962, 395 396)
Reworded in a different theoretical idiom: when beliefs and physiological states
contradict each other with respect to whether an emotional response is called for,
beliefs can trump physiology.
this should not be possible.
Impossible for Prinz:
*p
/no emotion
just any physiological change,
but a change of the sort that is normally sufficient to elicit an emotional
detection. Prinz holds that emotions are identical to these detections, and that
these detections sometimes register false positives. However, if emotions can be
suppressed by cognitive appraisals in spite of veridical detections, then there is
ions just are those detections.
This doubt remains even if detections are only rarely trumped, for if they are
ever trumped by beliefs then they cannot be identical to emotions. This is what I
call the New Adrenaline Objection (NAO).
4. Replies to the New Adrenaline Objection.
replies to the
original adrenaline objection defeat NAO.
third reply first,
since it is the most straightforward. That reply was that (P3) his view can
plausibly admit false detections, especially where these are caused by simulation
mechanisms parasitic on normal physiological feedback channels. This was
al appraisals with detections, and it is
not an undue strain on his model that detectors can activate falsely. If the
Akagi 6
detections are identical to emotions, then it is expected that the relevant
emotional states should obtain in case the physiological state-detectors a change,
whether or not they register veridically. However, NAO does not concern
detections (emotions) in the absence of sufficient physiological conditions. The
problem raised by NAO is that emotional states can fail to obtain even when the
proper physiological changes obtain and are detected. If emotions are identical to
detections, then this should not be possible.
replies also fail to address NAO because unlike the
motivated and the worries they express can be adapted to be responsive to NAO.
first objection (P1) was to question the assumption that subjects who
display different emotional behavior are actually in distinct emotional states.
n be reinterpreted as the worry that (P1′) the emotional
behavior
is not evidence of genuine
emotional episodes. Schachter and Singer used two measures to determine
emotional states: coded observation of participants by hidden experimenters, and
participant self-reports made after the departure of the confederate. Hidden
experimenters were uninformed as to the experimental condition of participants
(epinephrine-informed, epinephrine-ignorant, placebo, &c.) and counted
exuberant acts engaged in and initiated by study participants in the euphoria
scenario, or complaints in the anger scenario (Schachter & Singer 1962, 386
387). Observational
participants in the informed condition would display significantly fewer emotional
behaviors than participants in the ignorant condition, but Prinz observes that
participants in both emotion conditions self-reported having positive affect. Prinz
alleges that ignoring these self-reports is unprincipled, and that the disparity
undermi
(P1′)
reasons for questioning the self-report data in the anger condition are more
compelling than Prinz acknowledges. Schachter and Singer observe that study
participants were introductory psychology students who were compensated with
course credit, and were fearful that they would not be compensated if they
appeared to blow off the experiment.
they were alone with the [confederate], they hesitated to do so on
material (self-ratings of mood and questionnaire) that the
experimenter might see and only after the purposes of the
experiment had been revealed were many of these subjects willing
to admit to the experimenter that they had been irked or irritated.
(Schachter & Singer 1962, 391)
These after-the-fact testimonies cannot necessarily be trusted for the same reason
that the candidness of self-ratings was questioned in the first place, but they do
observational and self-reported measures. Prinz counters that if insincere displays
are a concern at all they should undermine the behavioral measures as well, for
Akagi 7
participants may have been trying to appease the confederate. However, the
preference for behavioral measures is motivated by feedback from study
participants, and not merely the theoretical zeal of the experimenters. Anyway,
since NAO does not acquire its force from comparisons between emotional
conditions, it would be enough that experimental measures were credible only in
the euphoria condition.
assumption that behaviors such as tossing paper basketballs, making and
throwing paper airplanes, and hula-hooping plausibly reflect genuinely positive
emotional responses.
This brings me to the second reason (P1′) fails to defeat NAO, which is
that the key comparison for NAO is
experiments, but between the epinephrine-informed and epinephrine-ignorant
conditions. Schachter and Singer propose that conflicting interests should lead us
to suspicion about the absolute value of self-reported affect
,
but claim that between-subjects differences may still be illuminating. Indeed, as
expected, the response trend
is that participants in the
epinephrine-ignorant condition rate themselves less happy than those in the
informed condition, although this comparison was not significant (p=.08). It
must be acknowledged that drawing any conclusions from the anger condition is
difficu
decisive. And as I noted before, NAO will succeed even if it is only borne out in
the euphoria condition, which (P1′) does not address.
Prinz defers in his methodological criticism of Schachter and Singer to a
which is critical of evidence supporting the Schachter-Singer theory of emotion,
does not address NAO since NAO does not presuppose the Schachter theory.
Reisenzein addresses results finding (1) the intensification of emotional reactions
through misattribution of extraneous cause, (2) the reduction of emotional
reactions through reduction of arousal, and (3) misattribution of emotional
arousal to a neutral source. None of these kinds of results includes NAO, which
concerns the reduction of emotional reactions through correct attribution of
increased arousal to an extraneous cause. The literature that concerns Schachter
ectly are two failed replications. The first, by
Marshall and Zimbardo (1979), compared misinformed subjects to a placebo
group, but did not include the informed/ignorant manipulation that is crucial to
NAO. The second failed replication, by Maslach (1979), had unrectified
methodological problems that are discussed by Reisenzein (1983, 248). There is a
1999), but this study also does not feature the informed/ignorant comparison
that is essential to NAO.
In fact, no attempted replication that I know of features the
psychologists have been interested in the soof emotion, these replications are better-designed than the Schachter and Singer
study. However, none of them bear on whether NAO is empirically supported. I
Akagi 8
acknowledge that there are some infelicitous methodological anomalies in
empirical basis for NAO, but to my knowledge there have been none.
(P2) concerned the assumption that identical
physiological states underlie different emotional displays. This response can be
disregarded, as the success of NAO does not depend upon there being identical
physiological conditions in the different emotion conditions. Freed from Schachter
an
theoretical baggage, we may include the questionnaire and the
r as eliciting conditions, even though they vary between
the
experiments. What concerns NAO is that these factors
do not vary across the informed/ignorant manipulation within each experiment.
More charitably, however,
second worry can be adapted to NAO as
the worry that (P2′) sufficiently similar physiological conditions did not obtain
between the epinephrine-ignorant and epinephrine-informed conditions. This
worry is difficult to assuage, since we do not know the sufficient or necessary
physiological conditions for inducing euphoria or anger, if there are any. Indeed,
although participants across conditions were in grossly similar states of arousal, it
would seem to beg the question against Prinz to insist that participants in the
informed condition were in states that should be sufficient to initiate emotional
episodes.
However the considerations raised by NAO are
s
account even if physiological preconditions did not obtain for epinephrineinformed participants, because any systematic differences in participant
physiologies between conditions must be attributed to different initial belief
states. Consider that even participants in a placebo condition displayed emotional
behaviors, and that they self-reported affective differences that trended more
intense than epinephrine-injected participants in the informed condition (though
these comparisons were not significant; 1962: 390, 392, 394). So the scenarios into
which the participants were put can plausibly be thought to constitute normally
sufficient eliciting conditions without epinephrine. The epinephrine injection
seems to have the effect of intensifying emotional behavior and self-ratings of
affect (despite anomalous absolute self, unless
participants are informed of the physiological effects of the injection, in which
case emotional behaviors and self-ratings are inhibited. So even if participants in
the epinephrine-informed condition fail to manifest normal physiological
concomitants of emotion, this failure must be attributed to a difference in their
beliefs. The incriminating beliefs in the Schachter and Singer study were acquired
via explicit testimony; they were linguistic, which is to say that their content was
conceptually-structured and therefore the knowledge gained by participants
consisted
in cognitive representations. The upshot is that even
if the informed condition was not such as to induce physiological states whose
would be because of a manipulation of cognitive representations.
As I mentioned above, Prinz allows
cognitive states can, through learning, come to function as eliciting conditions for
emotional responses when they activate pre-existing body-to-detection pathways.
Akagi 9
However, this appendix is unsatisfactory as a story about how cognitive states
can inhibit emotional responses when body-to-detection pathways are
independently activated.
version of their study, make perspicuous the complex dependence of emotional
states on cognitive states. This dependence on cognition undermines the
emotional episodes only by coming to replace perceptions in basic emotional
aetiological sequences. If cognitive states do not only elicit emotional episodes but
inhibit them, then
appears poorly motivated against an interactive account of emotions on which
physiology and appraisals both contribute to the manifestation of emotion in a
more complex manner.
5. Conclusion. I argued that a careful
results suggests an improved adrenaline objection, to the effect that there is
evidence for instances of physiological state-detections that fail to constitute
emotions. I addressed counternot indicate emotional episodes, and that physiological state-detections did not
obtain in the critical manipulation. Against the first it was conceded that there
condition. Against the second reply it was conceded that it could not be known
for certain without begging the question that sufficient physiological conditions
obtained, but that if they did not this fact must be attributed to cognitive
anyway. In any case, consideration of
identification of
appraisals with detections. Unfortunately, this identification was the centerpiece
of cognitive and somatic theories. If the foregoing
argument is
cannot quite be the right story about
the relations between bodies, minds, and emotions.
Akagi 10
REFERENCES
Damasio, Antonio R. 1994. Descartes' Error: Emotion, Reason and the Human
Brain. New York: Putnam.
Dretske, Fred. 1986. Misrepresentation. In R. Bogdan (Ed.), Belief: Form,
Content and Function (Oxford: Oxford University Press):17 36.
Lazarus, Richard S. 1984. "On the Primacy of Cognition." American Psychologist
39:124 129.
Marshall, G., and Philip G. Zimbardo. 1979. Affective Consequences of
Inadequately Explained Physiological Arousal. Journal of Personality and
Social Psychology 37:970 988.
Maslach, Christina. 1979. Negative Emotional B
Journal of Personality and Social Psychology 37:953 969.
Mezzacappa, Elizabeth S., Edward S. Katkin, and Stephen N. Palmer. 1999.
Cognition and Emotion 13:181 199.
Nussbaum, Martha C. 2001. Upheavals of Thought: The Intelligence of the
Emotions. Cambridge: Cambridge University Press.
Prinz, Jesse J. 2002. Furnishing the Mind: Concepts and Their Perceptual Basis.
Cambridge, MA: MIT Press.
---. 2004. Gut Reactions: A Perceptual Theory of Emotion. New York: Oxford
University Press.
Reisenzein, Rainer. 1983. The Schachter Theory of Emotion: Two Decades
Later. Psychological Bulletin 94(2):239 264.
Schachter, Stanley and Jerome E. Singer. 1962. Cognitive, Social, and
Physiological Determinants of Emotional State. Psychological Review
69(5):379 399.
Zajonc, Robert B. 1984. "On the Primacy of Affect." American Psychologist
39:117 123.