by Craig Delancey
Oxford University Press, 2002
Review by James Hitt on Apr 1st 2002
Engines: What Emotions Reveal About Mind and Artificial Intelligence puts
forth a position on the basic emotions.
Basic emotions constitute a small section of the emotions; they include
anger, fear, joy, and disgust, while excluding jealousy, guilt, wonder, and
hope. Basic emotions are panculturally
recognized in facial expressions of Homo
sapiens and the object of study for scores of neuroscientists experimenting
with mammalia. Basic emotions are primordial; they are
realized in neural systems old in evolutionary terms and contribute to the
realization of evolutionarily younger cognitive systems.
The last highlights DeLanceys
Hierarchical Model of Mind:
subcognitive affective systems underwrite the basic emotions, which are
independent of cognitive systems and components that are sometimes necessary
for cognitive systems. The contrast
between subcognitive and cognitive is a theme for many emotion theorists who
adopt, including DeLancey, the affect-program theory. DeLancey explains what that is and quotes approvingly from Paul
Ekman, a leading researcher on affect-theory: The term affect program refers to a mechanism that stores the patterns for
these complex organized responses, and which when set off directs their
. Affect programs coordinate complex bodily changes: such as facial
expressions and changes in the endocrine, respiratory, circulatory, or
musculoskeletal system. Besides
physiological responses, affect programs coordinate complex, although typical,
behavioral responses. The
affect-program theory, relying on the physiological and behavioral correlates,
is sufficient for identifying a basic emotion and distinguishing between basic
The affect-program theory
challenges a cognitive theory of the emotions.
On a cognitive theory, the emotions are essentially cognitive. The debate about whether the essential
features of (basic) emotions are cognitive is well represented in several books
for the lay reader. Joseph LeDouxs
book, The Emotional
Brain, is an excellent example.
In chapters 3-12, DeLancey explores
what the basic emotions reveal about philosophy of mind and artificial
intelligence. These chapters rely on
earlier arguments that purport to show basic emotions are subcognitive. In chapters 3-10, he surveys various issues
that are not always closely connected: interpretationalism, intentionality,
social constructivism, rationality, emoting for fiction, and phenomenal
consciousness. In chapter 11-12, he
addresses the theoretical underpinnings of artificial intelligence (AI). What is revealed about philosophy of mind
and AI is that they are both too cognitive.
Unable to account for subcognitive basic emotions, they fall short.
But it is arguable that DeLanceys
fundamental claim that every basic emotion is noncognitive is problematic. The claim seems to rely on an
equivocation. I will illustrate with a
simplified version of the argument against a cognitive account of the
emotions. First, a cognitivist claims
that (most) emotions are essentially cognitive* in that they require beliefs or
other propositional attitudes. But
basic emotions are not cognitive** in that they are non-linguistic, unavailable
for reporting, and processed quickly within the brain. So, basic emotions are not cognitive*. This naïve argument wears its fallacy on
its sleeve. The term cognitive is
used equivocally. DeLanceys position
is subtler, but the naïve argument nonetheless exhibits a persistent and
This is unfortunate. DeLancey has original and provocative ideas,
yet they emerge backstage when the main focus is on the inadequacies of
cognitivism for philosophy of mind and AI.
Three of his ideas immediately come to mind. First, DeLancey argues that basic emotions are motivational
states and not mere dispositions to behavior.
Second, DeLancey thinks that autonomy, rather than cognition, is the
central notion to be explicated in philosophy of mind and AI. Third, given DeLanceys understanding of
basic emotions, a passionate autonomous machine might be engineered. These three notions, if explored more fully,
promise to provide an alternative theory of emotion, rather than a theory of
basic emotions, and redirect some areas of the philosophy of mind and AI. In addition, these notions are seeds for
developing and articulating an entrancing discourse on the very feasibility for
constructing a passionate autonomous machine.
All the same, DeLanceys position
on the basic emotions faces an independent weakness, namely, that an
insufficient account of action is provided.
Human action requires agency.
When I run away from a dangerous brown bear, my behavior, the running,
is something that happens and something undertaken. Agency accounts for the latter; physiological states may well
account for the former. DeLancey takes
emotional behaviors to be actions. In
rejecting cognitivism as an account of the basic emotions, he foregoes a
standard account of action. For Donald
Davidson, a behavior is an action if it is caused by the reason for performing
the action. If an appropriate causal
connection between a reason and a behavior is absent, the behavior is not an
action. DeLancey, in presenting
counter-examples to Davidson, relies on the stipulation that all his examples
involving behavior are examples of actions.
If so, the person engaging in the behavior would, on a Davidsonian view,
have an appropriate belief-desire pair that constitutes her reason for
performing the said action. DeLancey
finds the belief-desire pair in each counter-example to be insufficient in
accounting for the agents behavior.
However, since the agent is performing an action, there is a
belief-desire pair that constitutes the reason for that action. On a Davidsonian view, showing that a
reason, that is, a particular belief-desire pair, is insufficient to account
for an action is perforce to show that that particular belief-desire pair is
not the reason.
I suspect DeLancey must in effect
be relying on an alternative account of action, perhaps one fashioned after his
notion of autonomy or motor program. At
most, DeLancey says that a motor program underwrites an emotional action. However, he provides no principled
distinction between programs that cause mere behaviors and programs that cause
actions (see p. 26), even though both are motor programs in the brain. What one demands before evaluating his
position on basic emotions is a principled way to distinguish between behaviors
that are actions and those that fail to be actions. A cognitive account, for all its difficulties, has the tools to
make available such a principled distinction, and it is not clear how the
affect-program theory can do that.
I applaud DeLancey for an ambitious
book. By joining together diverse
issues in philosophy of mind, our understanding of the emotions and philosophy
of mind is challenged and extended.
2002 James Hitt
Hitt is a Ph.D. candidate in philosophy at the Graduate School and
University Center, City University of New York. His is currently at work on a dissertation on the emotions.