What Emotions Really Are: The Problem of Psychological Categories (Volume 1997)

by Paul E. Griffiths

In this provocative contribution to the philosophy of science and mind, Paul E. Griffiths criticizes contemporary philosophy and psychology of emotion for failing to take in an evolutionary perspective and address current work in neurobiology and cognitive science. Reviewing the three current models of emotion, Griffiths points out their deficiencies and constructs a basis for future models that pay equal attention to biological fact and conceptual rigor.

"Griffiths has written a work of depth and clarity in an area of murky ambiguity, producing a much-needed standard at the border of science, philosophy, and psychology. . . . As he presents his case, offering a forthright critique of past and present theories, Griffiths touches on such issues as evolution, social construction, natural kinds (categories corresponding with real distinctions in nature), cognition, and moods. While addressing specialists, the book will reward general readers who apply themselves to its remarkably accessible style."— Library Journal

" What Emotions Really Are makes a strong claim to be one of the best books to have emerged on the subject of human emotion."—Ray Dolan, Nature

Genres: Philosophy, Science, Emotion, Psychology, Nonfiction

293 pages, Paperback

First published July 28, 1997

Reviews

[Alina · Rating 5 out of 5]

Griffith's take on emotion is one of a kind, or at least in the domain of philosophical works on emotion. Unlike virtually all other philosophers of emotion, Griffiths is highly informed by and explains in detail concepts in evolutionary biology. This work stands as an exemplar of a natural scientist who is interested, philosophically, in the question of what emotions really are; it does not come off as a work by a philosopher who dabbles in the sciences. This is also the only work in the philosophy of emotion which I know of which argues for an "eliminitivist" conclusion: the vernacular term "emotion" does not refer to any natural kind. I found various of his arguments in this book convincing and novel, and feel an urgency for philosophy of mind at large to take up the insights found therein. This is true in spite of that I do not agree with Griffith's overall conclusion and find him hypocritical in certain ways (i.e., arguments he makes to diagnose certain philosophers of conceptual ills equally apply to his own work.) Let me first summarize the chapters of this book, and then I'll rant/rave a bit at the end about it.

In chapter 1 "Introduction" Griffiths argues that virtually all philosophers of emotion think about emotion in terms of the propositional attitude (i.e., a mental attitude that is directed towards propositional content, like belief and desire.) He argues that this approach is doomed. This is because emotion inextricably involves bodily feeling, which is non-propositional in nature. Philosophers address this by proposing hybrid accounts of emotion, which define emotion in terms of a combination of feeling and propositional attitude; but this won't do.

His argument against this approach is as follows. The idea of the propositional attitude is formulated at the beginning as something that can be sufficiently studied by conceptual analysis (i.e., figuring out the necessary and sufficient conditions that something must meet in order to be classified under this concept.) This methodological attitude can work only if the semantics of a term is totally governed by the rules used by speakers in applying the term under consideration. But the semantics of natural kind terms violate that. The semantics of a natural kind term is at least partially determined by our standing in causal relations to the phenomena in the world thus tracked by the term. If we empirically discover that something we had thought is covered by the extension of a natural kind term does not belong there (e.g., whales as fish), we revise the intension of this term; so rule change is governed by our causal relations to the world, not by language games and convention alone. We shouldn't trust linguistic intuitions surrounding "emotion" to study emotion. We should look at what scientists have discovered about it.

Griffiths also stresses that we need not marry ourselves to a robust metaphysical realism of natural kinds in order to be on board with this picture of their semantics. Griffiths dives into the developmental psychological literature of how children acquire kind concepts. Children spontaneously create causal explanatory theories and cluster instances of a purported kind according to their having significant properties in light of these theories. When a child discovers a major counterexample or hole in their theory, they will update it. So causal theories of reference have some psychological truth. Griffiths will go more into his account of natural kinds when he reviews evolutionary biological literature in future chapters; biological and psychological natural kinds should be understood on the basis of how traits and organisms have actually historically evolved in lineages.

In chapter 2 "Philosophy and emotion" Griffiths goes more in depth into his charging philosophers of emotion with the crime of understanding emotion in terms of the propositional attitude. In chapter 3 "The psychoevolutionary approach to emotion" he reviews the historical biological literature on emotion, particular Darwin's and Ekman's views on this matter. In chapter 4 "Affect programs and emotion modules" Griffiths goes more in depth into affect program theory, as started by Ekman. He lays out the empirical evidence in favor of this theory and defends it from objections. On this theory, there are certain mental modules responsible for triggering affective activity (which has coordinated manifestations in physiology, psychology, and subjective feeling) in response to certain kinds of stimuli. These modules are evolved and universal across humans, and likely phylogenetic relatives. As modules, they are automated and mandatory; top-down cognition or volition cannot make a dent in the unfolding of these programs. Moreover, the learning and updating within these modules diverges from that of higher cognition; it can learn from subliminal, nonconscious stimuli pairing, for example, and from brute conditioning. It will not learn from conscious reasoning.

In chapter 5 "The higher cognitive emotions" Griffiths argues that there is a class of psychological states that are explained in terms of emotion, but which are not the same in kind as affect programs. He calls these higher cognitive emotions or social emotions. Examples include guilt, envy, and jealousy. These are not informationally encapsulated as affect programs are; they can be modified by volition and reasoning. He argues against the picture on which these social emotions are understood as the effects of affect (this argument passed over my head; I do not think there was a substantial argument. This will be a point I'll criticize below.) Rather, affect plays a role in the ontogeny of social emotions, but affect is not constitutive of them in any instance.

In chapter 6 "The social construction of emotion" Griffiths distinguishes between different theories and conceptions of "social construction" and argues that it is possible that certain social emotions are substantially socially constructed (i.e., we could not have these emotional responses without certain cultural conditions in place that shaped our upbringing.) He also argues against certain existentialist-influenced thinkers who want to say that as long as something is socially constructed, we have voluntary control over it; so when we claim that we're passive before a social emotion, like jealousy, this is a lie we tell ourselves. Griffiths points out that given how psychology works, social conditions can really condition into us automated emotion responses so that we are passive during social emotion episodes. (I also will criticize this point below.)

In chapters 7 "Natural kinds and theoretical concepts," 8 "Natural kinds in biology and psychology," and 9 "What emotions really are," Griffiths presents details from the evolutionary biological literature that are relevant to his key argument, which may roughly be represented as such: (1) Biological and psychological natural kinds should be demarcated on the basis of the same principles, (2) These principles include that instances of a kind at least need to share the same functional role and to be derived from the same evolutionary lineage, (3) Affect programs do not have the same functional role as higher cognitive emotions. Moreover, there are no evolutionarily-based analogues of these higher cognitive emotions in other species (which there are for affect programs); this counts against that cognitive emotions and affect programs could fall under the same natural kind, (4) "Emotion" refers to affect programs, higher cognitive emotions, and moods, (5) Emotion is not a natural kind (but only affect programs are.) These chapters are swimming in interesting details and careful explanations of certain concepts in evolutionary biology. I don't think all of them are relevant to Griffith's argument, but it was fun to read through nevertheless. If you don't want to read these chapters, that should be fine; the author presents the major points from this literature necessary to his argument in the first chapter.

Here are some of the details that I liked. Griffiths talks about "causal homeostatic mechanisms" as the justificatory basis for deeming a term as capturing a natural kind. A term must be able to let us accurately predict what future individuals fall under this kind; it must have explanatory power and be tested over time. Natural kind terms should not only aid good scientific theory; they should also explain why certain terms are deeply embedded in everyday use.

Griffiths reviews different principles invoked by evolutionary biologists for selecting natural kind terms, and argues for his champion one: cladistic clustering algorithms. Cladistics is roughly synonymous with studying life historically or genealogically; biological taxa are to be demarcated on the basis of shared historical lineage. The opposition to this is to think that biological natural kinds should be demarcated on the basis of "similarity" (e.g., all instances of a purported kind share some physical structure.) Griffiths shows that given the history of philosophizing about concepts, philosophers are prone to either taking this similarity approach in making sense of kinds. For biological kinds specifically, he urges that we need to look at phylogenetic facts too. Griffiths contrasts the idea of a "homologue" with an "analogue"; analogous traits may appear the same, but do not need to derive from the same phylogenetic lineage, whereas homologues do satisfy that historical criterion. For example, human limbs and bird wings are homologous, whereas the thermoregulatory capacities across humans and birds are not; these capacities arose independently in these two lineages. He argues that biological taxa demarcated on a phylogenetic/cladistic basis are explanatorily more powerful than those demarcated on other bases. For example, crocodiles are close to birds, and this would predict shared traits between them. This turns out to be true; both have complex vocalization patterns, nest building, and brood care, which are vernacularly most associated with birds. In contrast, other "reptiles" do not demonstrate these traits.

Griffiths also compares the levels of explanation approaches found in psychology and biology. Psychologists usually distinguish between three levels of explanation: (1) the implementation or neural level, (2) the information processing level, and (3) the task level. Evolutionary biologists often distinguish between four: (1) the physiological/anatomical level, (2) the cladistic level (the groups of lineages of organisms or traits, (3) the ecological level (what adaptive purpose a trait serves), and (4) the population dynamic level (generalizations of population genetics and evolutionary game theory, where traits are described by their fitness functions.) These levels are generally parallel between the fields, aside from that psychologists only talk about mental functions in the task level. Griffiths proposes that positing a mental function is explanatorily powerful only insofar as this function also satisfies cladistic/phylogenetic criterion. Functions that violate that criterion (e.g., "love" or reproduction as talked about across humans and AI) do not actually amount to natural kinds.

In the last chapter 10 "Coda: mood and emotion" Griffiths argues that what we colloquially refer to as moods sometimes overlaps with higher cognitive emotions and affect programs, but often do not. This reaffirms his point that "emotion" is not a natural kind.

Okie, so here are some of my thoughts. My criticism is that Griffiths fails to see how affect is necessary to what he calls higher cognitive emotion and mood. I believe there is a way to make emotion out as a natural kind, by which affect is understood as the functionally defined starting point, and then may be developed or lead to various kinds of consequences, where these are also functionally defined, so that the individual cognitive emotions and moods may all be understood as sharing this function. Here is an example of a detail that Griffiths does not consider, which could be used to proceed in constructing such an account. Language use and imagination can present distant parts of what we take to be reality. These can have the same functional role as the characteristic elicitors of affect, found in one's perceived environment. So higher cognition can be integrally involved in affect. Moreover, language and imagination can make it so that these parts of reality are relatively abstract/complex. For example, instead of the sudden loud sound of a gun firing in one's perceived environment, triggering the fear affect program, imagination/language use can present the possible state of affairs of one's beloved getting into a fatal accident (a case of imagining that may be triggered by the perceptual, actual state of affairs of one's beloved not having texted for a while).

Another puzzle piece necessary for my sort of account is that the sort of eliciting circumstances of affect programs can develop over time; we as infants perhaps start off with only sorts of circumstances that are found in the affect programs of our phylogenetic relatives (chimps, dogs), but over enculturation and maturation, we come to register other sorts of circumstances as occupying the same functional role, which updates details regarding that role.

So thoughts and imaginings and conversations can trigger affect, just as our perceptual environment can. Moreover, since these mental activities permit more abstract/complex contents than perception, the states of affairs thus delivered are relatively more "negotiable." The possible state of affairs that one's beloved has died can be negotiated with, since we do not know that it has not actually happened; we have not perceived it. If one first has the fear affect program triggered by this imagining, one can intervene in the unfolding affect by presenting to mind other facts that lead to convincing oneself that this possible state of affairs is very unlikely. Then that quells the fear.

This picture accommodates Griffith's point that higher cognitive emotions are subject to our voluntary control and reasoning in a way that basic affects are not. He uses this point to defend that the two apparent kinds of emotion are in fact distinct. But this need not be the case. The two apparent kinds can be understood as continuous and sharing the same function and evolutionary lineage once we bring to the table other facts about human psychology: e.g., that affect never occurs in a total vacuum, but we're always alert to what's going down in consciousness, and will be responsive to that; that much of what goes down in consciousness, which at first may show up with all the forcefulness of reality, may in fact end up to be flimsy and malleable, once we present counteracting or contrary parts of manifest reality via imagination and thought and conversation. These are all merely preliminary thoughts, of course.

As a whole, I'd highly recommend Griffiths to any student of philosophy interested in the emotions.

[Larry · Rating 4 out of 5]

Evolutionary psychological explanation should be historical, not just functional(ist), because if you don’t know how it evolved, you don’t know what trait evolved

[J.D. Steens · Rating 2 out of 5]

The title of this book is an overreach. In the end, after a vast amount of discussion, there's no good picture of what emotions "really" are.

Griffths starts out by criticizing "folklore" and everyday definitions, and states that we need to tap into the insights of Darwinian theory. From here he relies heavily on the "universal expressions" identified by Darwin as common to all humankind (anger, fear, disgust, sadness, joy, surprise), which Griffiths terms as primary. Griffiths then adds the "higher cognitive emotions" that involve appraisals, intention and beliefs (e.g., loyalty, commitment). Between these primary and complex poles, the author notes that there is a solid dose of social variability in how and when such emotions are expressed as each culture has its own "display rules."

All of this is clear enough I suppose. But beyond the few primary emotions listed by Darwin, there's only limited discussion about all of the other "emotions" and how they might fit into what Griffiths is trying to clarify in this book. Where does love fit? How about the social emotions that tie one to the group, or the deference that may tie one to one's leader? What is the distinction between a drive and an emotion and why is there that distinction? What about desire, pleasure and pain? Are some emotions more primary than others despite what Darwin said? Might fear be a more fundamental and generalized emotion than, say, disgust? And is there a relationship between these emotions; are, for example, fear and anger related? All in all, Griffiths does not propose any sort of theory that might tie these aspects together.

The more fundamental problem is that Griffiths, as many others, sees emotions as responses to outside stimuli (input is the outside trigger; output is the response). That leaves open the question about why one responds at all to outside triggers? Why, for example, do we fear? Why do we get angry? Why do we care enough to experience guilt and shame? What underlies these two responses and what are they if not emotions? By most accounts, love is an emotion yet strictly speaking, this is not a reaction but an action or at least an interaction. Love comes from the inside and radiates outward, looking for someone to love. The same could be said about the "inner promptings" (whatever these might be) that lead us to seek out friends and groups. Rather than viewing the self as a passive responder, an interactive theory of emotions has the self as an actor that either institutes actions or responds the way it does because it cares.

Perhaps we are too confined in the way we view emotions. Does "emote" mean to give off an overt, bodily expression as Darwin stated, or is it an inner, physiological need that prompts us to act or react the way we do, including, in some instances, giving off bodily expressions? Maybe it's more productive to see ourselves as filled, innately and universally, with energized needs that lead us to seek what we need from the world and to defend against what we don't need. In the absence of probing into these lines of questions, it seems doubtful that much can be said usefully about emotions.