Truly Enactive Emotion

Resisting the False Choice

Any adequate account of the emotions must accommodate the fact that emotions—including those of the most basic kind—exhibit intentionality as well as phenomenality. With this in mind, this article contends that a good place to start in developing such an account is by making certain adjustments to Prinz’s (2004) embodied appraisal theory (EAT) of emotions. If this suggestion turns out to be along the right lines, then it has important implications for the way in which we understand the nature and extent of the emotions, and thus, in turn, on how and where we might directly intervene on them.

Therefore what is at stake here should be of interest not only to those engaged in theoretical debates in philosophy, but also those in the practical business of psychiatry and clinical practice. These interests clearly intersect at this point. This is because developing successful therapies and deciding on appropriate interventions for dealing with emotional disorders depends on having an accurate understanding of the nature of the emotions to be dealt with. For anyone involved in such work it is vital to know if this or that theory of the emotions is true of the states of mind being targeted.

For example, some philosophers and psychologists regard emotions as contentful cognitive attitudes of some kind or other, such as evaluative judgements or appraisals (Armon-Jones, 1986; Nussbaum, 2001; Solomon, 1976). This is thought to exhaust their nature. Any such account trades under the “pure cognitivist” label.

There is no universally agreed understanding of what “cognitive” means (an issue that will raise its head again shortly). What unites the views just mentioned is, as Ratcliffe (2008) observes, that in “describing emotions as cognitive, [these] philosophers tend to mean at least that they are intentional states of some kind” (2008, p. 20).

Stressing this, pure cognitivism about the emotions is attractive because it satisfies “a deep intuition that emotions are meaningful. They … inform us about our relationship to the world, they embody our convictions, and they factor intelligibly into our decisions in life” (Prinz, 2004, p. 16). Certainly, if emotions are nothing but cognitive attitudes of some kind this would explain why they (or their essential components) have world-relating intentional properties. It is generally supposed that states of mind can only have such properties if they possess representational content.

Pure cognitivism comes in different forms depending on which notion of cognitive attitude is in play. For example, the attitudes in question might be thought to be explicitly formed propositional attitudes, such as beliefs or judgements. If so, they have the potential to be revised through dialogue that engages reasoning processes. For example, focusing on the necessary connection between having certain beliefs and experiencing shame, Prinz (2004) highlights some important implications of this version of pure cognitivism when it comes to thinking about how we might intervene on the emotions.

Not only can beliefs influence a state of shame, but they also seem to be obligatory for shame. If you do not believe you did anything wrong, you will not feel ashamed. Shame can be caused by beliefs and cured by beliefs. If you discover that your actions were beneficial rather than harmful, you can trade shame in for pride (Prinz, 2004, p. 83).

But pure cognitivism comes in other stripes too. For example, emotions might be belief-like states that are more basic than the sorts of explicit cognitive attitudes that can enter into our reasoning processes. It is easily conceivable that some cognitive attitudes might be immune to revision by rational means even though they possess representational content.

These differences matter practically. If pure cognitivism about the emotions is true, then clinicians will want to know which version is true. Or if there is a division of labour, and both are true to some extent, then they will want to know which account is true of specific kinds of emotions. For knowing this is of material importance for anyone tasked with devising treatments to deal with various emotional disorders. What we can know in advance is that if pure cognitivism is true, either in one or other of its more restricted forms or in some shared way, then the only way in principle to directly and completely alter someone’s emotions is to alter that individual’s cognitive attitudes. This will be so whatever practical difficulties happen to stand in the way of changing a certain individual’s cognitive attitude. Therapeutic interventions that target anything else, at best, offer indirect (and possibly only stopgap) measures for addressing emotional disorders.

Is pure cognitivism about the emotions true? Is it credible? It is certainly plausible that at least some emotions strongly depend on having certain cognitive attitudes. This can be accepted while leaving it open how best to understand the nature of the dependence relationship. Even acknowledging this, there are excellent reasons to doubt the truth of ambitious versions of pure cognitivism as providing an adequate theory of the emotions, whether separately or in coalition.

Pure cognitivist accounts of the emotions are woefully incomplete in that they overlook the importance of feelings. The recognized downside of purely cognitivist accounts of the emotions is they aren’t capable of telling the whole story about emotions. For it is easy enough to imagine the relevant cognitions taking place in disembodied, entirely “cold,” “detached” and, wholly, “unemotional” ways.

This observation is especially pertinent even if we consider quite ordinary cases. Familiarly, even when our emotions are stirred up by having certain beliefs or judgements, the associated feelings relating to such emotions can outlast the changes in those beliefs and judgements. Thus my wife’s seething anger and feelings of outrage at what she takes to be my transgression might not immediately subside upon discovering me innocent of that of which I am accused. While, as a consequence of this discovery, she may no longer direct anger at me (and certainly not with justification) it does not follow that her feelings of anger entirely dissipate, and in such circumstances, it is the having of such feelings that seems sufficient for her being in a continued emotional state of anger.

Considerations of this kind can make the somatic feeling theory of emotions appear attractive. According to that theory, emotions are just feelings of bodily changes as they occur (James, 1884). This basic Jamesian idea has been revamped and updated in the somatic perception theory advocated by Damasio (1994). Accordingly, James’ original account is revised to allow for emotions to occur even without the relevant bodily changes, just in case the relevant brain activity that monitors the bodily changes is present. Hence, “as if” feelings triggered by central processes in the brain suffice for having certain emotional experiences. Such theories are problematic in that they “have little to say about the processes by which external stimuli are evaluated for ecological and social significance” (Hill, 2009, p. 199). Basically, in reducing emotions to bodily feelings or perceptual states exclusively targeting such, “somatic theories have trouble explaining what it is for an emotion to have an intentional object or target” (Hill, 2009, p. 200). They lack appropriate reach. Such theories lean too far in the nonintellectualist direction.

It seems that if we are to have an adequate account of what is essential to being an emotion, we must resist the false choice between pure cognitive theories and pure bodily feeling theories. Apparently, what is really needed is an account of “how emotions can be sophisticated cognitive states and, at the same time, [emphasis added] have bodily feelings as a major component” (Ratcliffe, 2008, p. 17). Prima facie, this looks difficult to achieve. Ratcliffe’s diagnosis is that we will only do so by overhauling some deeply entrenched constraining assumptions: “Central to this overhaul is the abandonment of the distinction between cognition and affect” (Ratcliffe, 2008, p. 17). I concur that this is broadly right—and finding a way to adequately revise our assumptions is, if anything is, a job for a philosopher.

Embodied Appraisal Theory: Putting Humpty Together Again?

Enter Prinz’s embodied appraisal theory (EAT)! As the name suggests, EAT is meant to be an embodied—not a cognitivist—theory of emotion. It says that emotions are noncognitive, embodied appraisals. But, importantly, EAT only promotes noncognitivism in the special sense that it recognizes that emotional engagements involve brain regions other than those in prefrontal areas. For Prinz, this is sufficient for emotions to count as embodied, as opposed to cognitive. This is because he holds that “a state is cognitive just in case it includes representations under the control of structures in executive systems, which, in mammals, are found in the prefrontal cortex” (Prinz, 2004, p. 47). Prinz’s noncognitivism is therefore motivated solely by the fact that, without engaging the neo-cortex at all, “the amygdala can orchestrate the full suite of bodily and behavioural responses that are associated with fear” (Prinz, 2004, p. 34). His assumption, of course, is that fear is not special in this respect and that the other emotions are also free from control of prefrontal areas.

Yet, despite being noncognitive in this sense, EAT understands emotions to be essentially representational in nature (and so emotions qualify as “cognitive” in the sense discussed in the previous section). Prinz’s reasoning is as follows: “Appraisal theories claim that emotions necessarily comprise representations [emphasis added] of organism–environment relations with respect to well-being” (Prinz, 2004, p. 52). With this in mind, he believes that “to show that emotions are appraisals, one must first establish that they are mental representations” (Prinz, 2004, p. 52). In what follows I attempt to resist this ruling.

Prinz, by contrast, tries to meet it by adopting what is probably the most popular theory of mental representation on the contemporary scene—teleosemantics. Accordingly, he holds that:

a mental representation is a mental state that is reliably caused by something and has been set in place by learning or evolution to detect that thing … a mental representation is a mental state that has been set up to be set off by something. (Prinz, 2004, p. 54)

Against this backdrop, in promoting EAT he makes the additional move of assuming that the representations required for embodied appraisals are “inextricably bound up with states that are involved in the detection of bodily changes” (Prinz, 2004, p. 52). Although this is meant to capture the core insight of the somatic feeling and perception theories mentioned earlier, EAT differs importantly from such theories because it also claims that the relevant registrations and detections of bodily changes must be in the service of representing organism–environment relations that matter to the organism’s well-being.

The reason is simple. Prinz insists that emotions must “detect something more than the vicissitudes of vasculature. Otherwise, they would confer no survival advantage” (2004, p. 60). It is by having intentional directedness that aims at environmental situations of special importance to the organism that emotions can play their distinctive roles in guiding our activity. Thus “Emotions promote behavioural responses. We [are meant to] run when we are afraid [of something dangerous]. If emotions represented bodily changes this would be unintelligible. We should flee when our hearts race” (Prinz, 2004, p. 59).

Prinz, borrowing from Lazarus, holds that the many and various worldly events and situations that elicit specific types of emotions, such as sadness or fear, can be classified as representing core relational themes, such as loss or danger. Lazarus (1993) tells us “Each emotion arises from a different plot or story about relationships between a person and the environment; feeling angry has its own special scenario, and so does feeling anxious, guilty, ashamed, sad, proud, and so forth” (1993, p. 12). EAT holds that these themes are the foci of embodied appraisals. Accordingly, Prinz holds that emotions do not have ordinary intentional objects, such as particular objects or people, as their proper targets—except incidentally or upon recalibration when the emotions take on additional representational objects.

Prinz holds that the core relational themes are wholly objective and mind-independent; they are not to be understood as any kind of response-dependent properties. It is disputable that this is the best way to understand their nature (see D’Arms, 2008). For reasons of space I will not examine this question in this article. It suffices to note that it is not necessary to follow Prinz’s lead on this issue.

Prinz defines the object of a given emotion as the environmental situation (actual or imagined) that elicits it. Importantly, he distinguishes a formal object of an emotion (which is the property in virtue of which an event is meant to elicit an emotion) and a particular object of emotion, which is the eliciting event itself (Prinz, 2004, p. 62). With these distinctions in place he acknowledges an important ambiguity in our talk of “aboutness,” remarking that:

Saying that my sadness is about the death [of a loved one] does not mean that my sadness represents the death; rather it means that the death is what has caused me to become sad. I can continue to think about the death after my sadness subsides, and I can continue to be sad after my thoughts of the death subside. (Prinz, 2004, p. 62)

In accord with teleosemantics, the idea that emotions represent core relational themes translates into the idea that we are “set up to be set off” by events or situations having such thematic properties, since these kinds of situations will have mattered to our ancestors. Assuming things have not changed dramatically for us, not only do such themes continue to have significance for our well-being, it follows that, when conditions are right, we can be made aware of events relating to such themes by registering changes in our bodies and thus representing things truly, as they are (Prinz, 2004, p. 68; see also pp. 66–67). And so, we come to EAT’s big reconciling idea: “Each emotion is both [emphasis added] an internal body monitor and [emphasis added] a detector of dangers, threats, losses, or other matters of concern. Emotions are gut reactions; they use our bodies to tell us how we are faring in the world” (Prinz, 2004, p. 69).

There is nothing ad hoc about this feature of the proposal from the perspective of teleosemantics. According to that theory of mental representational capacities, “mental representations often work this way. They track real contents via more superficial nominal contents” (Prinz, 2004, p. 68). In her frugality Mother Nature supplies us with less than perfect appearance-tracking detectors, not essence-tracking detectors. These are good enough to enable us to get by (at least compared to the means used by our ancestral competitors).

A nice feature of EAT’s account of the origins of basic emotional experiences as a way of responding to core relational themes—one that is especially relevant when thinking about emotional disorders—is that it enables us to make sense of misaligned emotional experiences. Hence,

Sometimes we are sad when there has not been any loss. This might occur under the influence of certain drugs (e.g. alcohol), while listening to music, or even while making a sad facial expression. Some cases of clinical depression may involve chronic sadness without any loss. (Prinz, 2004, p. 64)

Radical Enactivism: Tweaking EAT

Prima facie EAT has the potential to bring the cognitive and feeling aspects of emotions together again in a unified account. It rejects the conceptualization hypothesis, assuming that judgements and appraisals do not require the possession of concepts. It also rejects the disembodiment hypothesis, assuming that emotions cannot occur without at least some somatic component. But it accepts the appraisal hypothesis, assuming that emotions involve representations of the organism–environment relation. With regard to the latter, EAT follows Arnold (1960) in stressing that “to appraise something is to see it as affecting oneself in some way that matters” (1960, p. 171).

EAT has promise, but it is problematic as it stands. One problem with EAT—the one that I am concerned with in this article—is that it depends on teleosemantics, which I take to be a failed naturalized theory of mental content. If so, that’s bad news for EAT. With this in mind, my game plan is to show how the central idea of EAT (or something in the vicinity) can still be made to work even if we make an important technical tweak—a tweak that allows us to address a weakness that would otherwise leave EAT vulnerable to objections to the foundation of the theory. This can be easily done if we recast EAT as an account of the emotions based on teleosemiotics, not teleosemantics. EAT can be improved if we abandon its semantic gloss; if we let go of the idea that emotions represent situations in truth-evaluable ways and that all appraisals must be representational in this sense. This requires giving up the idea, for example, that in the right conditions my anger represents truly that there has been a demeaning offence against me and mine. Rather, the guiding insight of EAT is revised to say that we are set up to be set off by core relational themes by responding to these in distinctive ways that implicate experiences of bodily feelings. As a result, to respond in such ways to specific kinds of situations is entirely natural but those ways of responding do not, in and of themselves, “say” anything about how things stand with the world.

For reasons of space, it is not possible to rehearse the detailed arguments for thinking that we ought to trade in teleosemantics for teleosemiotics here (these can be found in Hutto, 2008, Chapter 3; Hutto, 2011; Hutto & Myin, in press). In the briefest of sketches, this antirepresentationalist turn in thinking about basic intentionality is motivated by the following sorts of consideration. To qualify as representational in a properly semantic and contentful way, a mental state must play a special kind of role in a larger cognitive economy. Crudely, it must, so to speak, have the function of saying or indicating that things stand thus and so, and being consumed by other systems because of what it says or indicates. Only if it does this will an internal state or structure meet Ramsey’s (2007) job description challenge. But it is plausible that many of the states (or ensembles of states) of organisms that enable basic cognition, including emotional responding, are merely (a) reliably caused by (or nomically depend upon) the occurrence of certain external features, (b) disposed to produce certain effects (under specific conditions), and (c) have been selected because of their propensities for (a) and (b).

Yet states or structures that only possess properties (a)–(c) fail to meet the job description challenge. They fail to qualify as truly representational mental states having the proper function of saying “things stand thus and so,” rather, they only have the proper function of guiding a system’s responses with respect to specific kinds of worldly offerings.

Additional and related arguments put forward, for example by Fodor (2008), purport to show that Darwinian-inspired theories of content can, at best, account for states of mind exhibiting intentional (with a “t”) directedness, but they flounder when it comes to account for those exhibiting intensionality (with an “s”)—and the latter are required for having truth-evaluable thoughts.

Indeed, even those who were initially friendly and optimistic about the prospects of teleosemantic theories have come to doubt that such theories have any chance of success in achieving their stated aims of providing properly semantic theories of mental content. Godfrey-Smith (2006) provides an astute assessment: “there is a growing suspicion that we have been looking for the wrong kind of theory, in some big sense. Naturalistic treatments of semantic properties have somehow lost proper contact with the phenomena” (2006, p. 42). Nevertheless, he also acknowledges that the driving idea behind teleosemantics—that evolved structures can have a kind of “specificity” or “directedness”—is essentially correct: “there is an important kind of natural involvement relation that is picked out by selection-based concepts of function. But this relation is found in many cases that do not involve representation or anything close to it” (2006, p. 60).

In short, it is becoming clear that teleosemantic accounts lack the right resources for naturalizing properly semantic properties, such as truth and reference, using purely biological tools. Thankfully, providing theories of mental content is also unnecessary in order to understand basic forms of intentional directedness. Nothing important to EAT is lost if in the place of teleosemantics we put teleosemiotics. Teleosemiotics is (basically) teleosemantics without the semantic ambitions. Teleosemiotics borrows what is best from teleosemantics to provide a content-free, naturalistic, account of the determinate kinds of intentional directedness that organisms exhibit towards aspects of their environments. Yet unlike teleosemantics, it does not seek to understand the most basic forms of directedness in semantic, contentful or representational terms. It holds that the biologically basic modes of organismic responding are not content involving, where content is understood in terms of reference or truth conditions.

Modifying EAT by swapping its reliance on teleosemantics for teleosemiotics confers another advantage in addition to avoidance of potential theoretical bankruptcy. Colombetti and Thompson (2008) complain of EAT that, despite the promise of its name, it is not a genuine embodied account of emotion. This is because by its own lights “the body still plays the role of an objective concomitant of emotion. It is the appraisal component that is seen to be in charge of providing personal significance” (Colombetti & Thompson, 2008, pp. 54–55).

Their chief objection to EAT is that, through its commitment to neural representationalism, it continues to promote separatist thinking: “it holds onto a disembodied conception of cognition (appraisal), and takes cognition so understood to be the source of the meaning of an emotion” (Colombetti & Thompson, 2008, p. 55). Against this, they press for the more radical idea that:

cognition is a form of embodied action … the enactive approach implies that we need to move beyond the head/body and subjective/objective dichotomies that characterize much of emotion theory. Appraisal is not a cognitive process of subjective evaluation “in the head” [emphasis added] and arousal and behavior are not objective bodily concomitants of emotion. Rather, bodily events are constitutive of appraisal, [emphasis added] both structurally and phenomenologically. (Colombetti & Thompson, 2008, pp. 56–58)

Original variant EAT assumes that the representational vehicles of emotional content are wholly embrained—that they reside somewhere in the amygdala. This is wholly in line with standard thinking about the nature of mental representations. For it is widely supposed that we can distinguish the vehicles and contents of mental representations by appeal to the sentential model. Neural vehicles are said to bear mental contents in exactly the same way that sentences bear linguistic contents. As such there are important differences between vehicles and contents. On the one hand, it makes no sense to say that the entities and events represented, say, in a story (its contents) reside in a book. Drawing on this very example, Dretske (1995) tells that no one expects to find damsels in distress, brave knights, or feats of daring-do in books. Still, it makes perfect sense to think that sentences describing those events (the vehicles of those very contents) can be found in specific locations in a book. By the same token, it is assumed that neural vehicles of representational content are locatable, in principle, in specific places in brains.

Now, things look quite different for a modified EAT—one that swaps teleosemantics for teleosemiotics. It denies that we should think of the bodily basis of emotional responses as content-carrying vehicles, full stop. This is because an EAT that is modified along the proposed lines rejects the idea that basic intentionality is a matter of responding to or consuming contents and, hence, it rejects all vestiges of the sentential model upon which the distinction between representational vehicles and contents rests.

In this respect, the radical brand enactivism on offer here goes further than other enactivist accounts of the emotions in thoroughly rejecting talk of contents and meanings at the level of basic responding. It rejects the idea that organismic responses relevant to having emotions are responses that create, carry, and consume meanings. By comparison, Thompson (2007) tells us that in cases of emotional responding:

sensory stimuli … induce the construction by nonlinear dynamics of an activity pattern in the form of large-scale spatial pattern of coherent oscillatory activity. This pattern is not a representation of the stimulus but an endogenously generated response triggered by the sensory perturbation, a response that creates and carries the meaning [emphasis added] of the stimulus for the animal. This meaning reflects the individual organism’s history, state of expectancy, and environmental context. (2007, p. 368)

Similarly, Colombetti stresses the “meaning-generating” (2010, p. 146, original emphasis) role of the body and speaks of the nonneural body as a “vehicle of meaning” (2010, p. 147). Thus, we are told that “meaning is generated within the system for the system itself – that is, it is generated and at the same time consumed by the system” (2010, p. 148, original emphasis). Although it is emphasized that the meaning in question is always relational and codetermined by the organism and its environment, it is not clear how to make sense of the metaphors of production and consumption or how to understand what is consumed or produced. Taken at face value, they suggest additional, special processes that are somehow distinct from the way an organism responds to the worldly offerings in intentionally directed and, perhaps, emotionally charged ways. But if talk of meaning and vehicles of meaning does not imply distinct and additional processes then it is not clear what work it does in these accounts.

Another difference is that Colombetti and Thompson stress that their type of enactivism offers a way to truly close the imagined gap between cognition proper, on the one hand, and mere bodily and affective activity, on the other. This is achieved by adopting a quite liberal understanding of the nature of cognition. Thus Thompson proposes that

cognitive interactions are those in which sensory responses guide action and actions have consequences for subsequent sensory stimulation, subject to the constraint that the system maintains its viability. “Sensory response” and “action” are taken broadly to include, for example, a bacterium’s ability to sense the concentration of sucrose in its immediate environment and to move itself accordingly. (Thompson, 2007, p. 125)

Talk of “cognition,” “interpretation,” “sense-making,” “understanding” and even “emoting” in describing the responses of simple living systems is misplaced and misleading. To some extent enactivists admit this. Indeed, on this score, when they giveth they then quickly taketh away. Thus Varela, Thompson, and Rosch (1991) note that they use the terms “significance” and “relevance” advisedly in speaking of simply creatures enactively bringing forth their worlds. Consequently, even though these authors assume that some kind of interpretation takes place in such cases, they confess that “this interpretation is a far cry from the kinds of interpretation that depend on experience” (1991, p. 156). I am inclined to take a harder line: this is no kind of interpretation. The simplest life forms are capable of an intentionally directed responding of a kind that when suitably augmented provides a necessary platform for cognition, interpretation, understanding, sense-making and emoting.

Still, there seems no reason to rule out that such activity constitutes a basic kind of intentional directedness—intentional directedness of a sort that is shared with more sophisticated forms of life that possess more elaborate states of mind. The task is to explain how such basic states of mind become possible when other proper supports are in place in the form of the right kinds of interactive engagements and practices. ¹

What can be agreed is the more modest claim that basic interest-driven ways of responding provide the right platform for understanding how emotions can be both intentionally directed attitudes yet also embodied feelings. Certain organisms are not only set up so that they are intentionally directed at situations that bear on their well-being; in some cases, their ways of responding constitute phenomenally charged ways of responding that are properly emotional, despite being wholly nonrepresentational. By modifying EAT in the ways proposed—by letting go of its commitment to semantic representationalism—we come by a truly enactive account of emotion.

To respond emotionally is to respond in ways that have a distinctive phenomenal character. But it also assumes that to respond in such ways is identical with, and to be understood in terms of, concrete patterns of environmentally situated organismic activity; nothing more, nothing less. Phenomenally charged experiencing simply equates to the way in which certain creatures are disposed to respond to a range of worldly offerings. But understanding what it is that sets up a creature to respond in these ways is a different matter. For this reason, even on the deflated, nonsemantic understanding of intentional directedness offered by teleosemiotics, the intentional aspects of emotion do not reduce to its phenomenal aspects (nor vice versa).

While this approach leaves open the task of describing the phenomenal character of particular states of mind, we should refuse to conceive of the felt quality of anger or fear as something over and above, and thus logically distinct from, organismic ways of responding. This is the wont of qualia-philes. For radical enactivists, the phenomenal character of an experience is not identified with, or determined by, extra ingredients over and above the dynamic responses of organisms. Instead they are identified with specific, concrete activities of sentient beings—extended ways of responding, reacting, acting and interacting. To understand how things seem or feel to us just is to understand this sort of activity, though—to be sure—it does follow that even an accurate description of such activity will convey an insight into the relevant phenomenology.

Despite the differences between a radical enactivist approach to emotion and the other enactivist accounts mentioned before, enactivists of all stripes agree that the rejection of representationalism entails that the kind of organismic activity that constitutes any emotional episode will be a response of the whole organism (see Thompson, 2007, p. 363). Hence, if we construe EAT as a response-based, as opposed to a standard representational, account of basic emotional engagements, it is wholly amenable to an enactivist (and, potentially, a fully embodied and not merely embrained) rendering. By its lights, there is surely no need to assume that emotional appraisals are essentially representational and hence solely “heady” affairs. It also means that we can intervene on the emotions directly without focusing our efforts solely on the brain because emotional responding is extensive. Emotions are ways in which organisms interact with their environments that involve, but are not exclusively isolated to, what goes on in their brains.