Emotions or Emotional Feelings?

Commentary on The Brain and Emotion by Edmund T. Rolls
for the BBS Multiple Book Review
To appear in BBS, April 2000, Vol. 23, No. 2.

Murat Aydede

The University of Chicago
Department of Philosophy
1010 East 59th Street
Chicago, IL 60637, U.S.A.

EMAIL: m-aydede@uchicago.edu
URL: http://humanities.uchicago.edu/faculty/aydede/

July, 1999

ABSTRACT. I criticize Rolls’s account of what makes emotional states conscious.

It turns out that Rolls’s answer to Nagel’s (1974) question, "What is it like to be a bat?" is brusque: there is nothing it is like to be a bat . . . provided that bats don’t have a linguistically structured internal representational system that enables them to think about their first-order thoughts which are also linguistically structured. For phenomenal consciousness, a properly functioning system of higher-order linguistic thought (HOLT) is necessary (Rolls 1998, p. 262). By this criterion, not only bats, but also a great portion of the animal kingdom, perhaps all animal species except humans, turn out to lack phenomenal consciousness. Indeed, even human babies, and perhaps infants before the early stages of acquiring their first language, are likely to lack such consciousness, if one considers the level of conceptual sophistication required by the HOLT hypothesis. In order to have a higher-order thought, one needs to have the concept of a thought in addition to the (linguistically structured) representational resources to articulate the conceptual content of the lower-order thought. Indeed, Rolls believes (p. 262) that phenomenal consciousness may be quite a late arrival in the history of evolution of the mind/brain: certainly much later than the ability to think (in a linguistically structured internal medium), because it requires the ability to think not only about one’s physical environment, but also about one’s own thoughts. Many would take such consequences as a reductio of Rolls’s thesis about consciousness; for surely the thesis seems to overintellectualize phenomenal consciousness.

But Rolls does not deny that animals have emotions, only that they have emotional feelings. The latter are conscious, whereas the former are not. There is nothing it is like to be in an emotional state unless one is endowed with a HOLT system. But why does Rolls feel he must embrace this startling and implausible conclusion? After telling us that emotions are brain states caused by positively or negatively reinforcing stimuli, including changes in such stimuli, Rolls offers a fascinating tour of the physiology of reward and motivation in which he makes quite elaborate and specific proposals as to where and how in the brain such states are likely to occur (primarily in the amygdala and orbitofrontal cortex). But Rolls’ actual "definition" of emotion is only part of the story. Later (especially in sects. 4.6–4.9, 9.3, 10.3, among others), he implicitly supplements this definition by telling us that emotions are also the indirect causes of certain types of motivated behavior. In particular, they are (in humans) inputs to two brain systems that decide on the behavioral output by computing the reward value of each behavioral choice against its odds before outputting to motor areas. One system, which humans share with other primates, consists of the basal ganglia and its structures (including, perhaps, the inferior temporal visual cortex — cf. p. 285). The output of this system is processed in the premotor cortex and then fed into the motor system before the actual ensuing behavioral response. Rolls calls the behavior elicited by this route "implicit behavior." The other system, which is perhaps specific to humans, is the language cortex, which receives the outputs of amygdala and orbitofrontal cortex, processes them in a syntactically structured symbolic medium before initiating action through the cortical motor and planning areas. Behavior from this route is dubbed "explicit behavior". (In addition to these two, there are also reflex circuits between relatively unprocessed sensory inputs and motor reactions.)

So the core of Rolls’s account is a straightforward functionalist (in fact, psychofunctionalist — see Block 1980) characterization of emotional states. Hence we can explicitly rewrite Rolls’s actual definition thus: emotional states are those states (mainly realized in the amygdala and orbitofrontal cortex) that play the above-specified type of functional/causal role (to be specified more fully and explicitly of course) in the central neural economy of the brain. But then Rolls asks: Why should there be anything it is like to be in such states? Why should such states feel like anything at all? (I am of course delighted to see a scientist who thinks that he should have an answer to this question. So I truly applaud Rolls’s attempt to give an answer — albeit a tentative and cautious one.) Feeling that his account of emotions seems to leave out the qualia of emotions, Rolls proposes the HOLT hypothesis.

But I am puzzled by why Rolls thinks that this helps. Before explaining my misgivings, a few clarifying remarks (or exploratory speculations) about the hypothesis are in order. As Rolls points out, the HOLT account closely resembles Rosenthal’s HOT theory of state consciousness, though it adds the requirement that thoughts are realized in a syntactically structured (language-like) representational medium, or "mentalese," as per the Language of Thought Hypothesis (LOTH — for a presentation of which, see Aydede 1998). After distinguishing between first- and higher-order thoughts (all realized in mentalese), Rolls insists that second-order thoughts (i.e., thoughts about first-order thoughts) are required for emotions to become conscious, that is, to turn them into emotional feelings. This is a bit puzzling, because, as characterized by Rolls, emotional states are obviously not thoughts realized in mentalese. Rolls sometimes talks about the firing of a bunch of (specific) neurons in the orbitofrontal cortex as a representation of the reward value of, say, a particular taste — intuitively what we might otherwise call, its pleasantness, or the lack thereof. But, as far as I can tell, these representations are not privy to the computations of the explicit linguistic system. (However, see Sect. 4.6.3, where Rolls seems to suggests, somewhat confusingly, that the computation of reward value in the linguistic system might depend on activity in the orbitofrontal cortex and amygdala, which are not identified by Rolls as sites of the higher linguistic cortex.) So it is not clear how these representations are supposed to be made conscious by shining the light of higher-order thought on them.

Perhaps all Rolls means to say is this: in order for emotional states (say, orbitofrontal representations of reward value) to become conscious they must be the target of a first-order thought; nothing more is required. On this interpretation, all that the organism needs to do is to think about such representations — perhaps something like the internal version of "Oh, this is pleasant" where "pleasant" is represented by a mentalese predicate, and "this" is a Mentalese quasi-demonstrative referring to . . . what? It cannot be the orbitofrontal representation, for that is not pleasant. What is pleasant is presumably the taste, which is processed, according to Rolls, independently of and prior to its reward value, in the primary taste cortex. So perhaps, it refers to a purely sensory (i.e., affectively neutral) representation of that particular taste, say in the primary taste cortex. But this does not seem right either. How could an affectively neutral sensation be pleasant? But then what is the first-order thought supposed to be about? What these questions/reflections seem to indicate is that if a version of HOT theory is claimed to account for emotional feelings, then this account is absent from Rolls’ book.

But suppose that such an account were provided. Surely it would be a causal/functional one, in that the postulated higher-order states will be brain states causally connected, perhaps in quite complicated ways, to the sensory and affective states realized in the relevant brain sites. So, for instance, suppose whenever I am in those emotional states, I am causally prompted to have thoughts (brain states realized up in the linguistic cortex) about my emotional states (just further brain states down in the orbitofrontal cortex and amygdala) that — voilà — transform the latter into emotional feelings, full-blown phenomenal states in all their glory. How is this supposed to be less mysterious? Why should there be anything it is like to be in a state playing this (more sophisticated) causal/functional role? Contrary to Rolls, as things stand, I do not see any advance here.

Nevertheless, I am no pessimist or mysterian about phenomenal consciousness. On the contrary, I believe that the mystery can probably be solved by more or less the same naturalistic or scientific methods that Rolls so skillfully employs throughout the book in uncovering some of the brain mechanisms of emotions. I even believe that he is right in thinking that at some point in our analysis of phenomenal consciousness we need to bring in some form of higher-order mental state theory (this could be a HOT or HO-Perception account in the way introspection has been traditionally conceived, i.e., as a kind of inner sense). But to solve the mystery of affective/emotive qualia will require at least two things. First, we need to account for such qualia not representationally, but rather in purely psychofunctional terms, namely as ways of processing incoming sensory information en route to setting behavioral parameters. I believe Rolls’ research gives substantial support to this kind of approach. Second, we need to tell a naturalistic story about our peculiar first-person epistemic access to them, a story which does justice to the subjectivity of the mental. This is where going higher-order is likely to play an essential role, especially when we bring in some of the resources and peculiarities of indexical and reflexive reference to such a account. For exploration of some of these themes, see Dretske (1995), Lycan (1996), Rey (1997), Tye (1995), and Aydede (in preparation).

Acknowledgment: Many thanks to Philip Robbins for his comments and corrections.


Aydede, Murat (1998). The language of thought: state of the art. Available at: http://humanities.uchicago.edu/faculty/aydede/LOTH.SEP.html.
A shorter version appeared in the electronic Stanford Encyclopedia of Philosophy, edited by E. Zalta:

Aydede, Murat (in preparation). Naturalism, qualia and pain. The University of Chicago, available at http://humanities.uchicago.edu/faculty/aydede/pain.pdf

Block, Ned (1980). What is functionalism? In Readings in Philosophy of Psychology, Vol.1, edited by N. Block, Cambridge, MA: Harvard University Press.

Dretske, Fred (1995). Naturalizing the Mind. Cambridge, MA: MIT Press.

Lycan, William G. (1996). Consciousness and Experience. Cambridge, MA: MIT Press.

Rey, Georges (1997). Contemporary Philosophy of Mind. Oxford, UK: Blackwell.

Rolls, Edmund T. (1998). The Brain and Emotion. Oxford, UK:Oxford University Press.

Nagel, Thomas (1974). What is it like to be a bat? In Philosophical Review 83:4, 435–450.

Tye, Michael (1996). Ten Problems of Consciousness. Cambridge, MA: MIT Press.