Department of Cognitive Science
University of California San Diego
johnson@cogsci.ucsd.edu

Language is a highly complex process with a variety of interacting components. This paper will focus on one such component, the control of affective expression, examining its role in language development, and its possible evolutionary origins based on a comparative analysis of its rudiments in nonhuman primates.

The impact of this component is most starkly evident in very early language development when the infant, who has been babbling for some months, begins to use its first words (~1-1.5 years old). During this period, there are two apparent sources for these utterances, whose operation, I will argue, are inter-related. One is the (relatively faithful) imitation of externally generated sounds, such as the vocalizations of others. Much has been written about this sort of mimicry (e.g. Meltzoff 1988; Donald 1993) which clearly plays a vital role in the development of symbolic reference. For the purpose of this paper, however, I will focus more on the other source – the imitation, or modified repetition, of internally generated sounds. These may include "inventions" of the infant’s articulatory system or utterances incorporated through the above-mentioned external imitation process. In either case, it is the manner in which such repetitions are modified that reveals the operation of the affect control mechanism.

These modifications may be best understood by approaching early language development from a Vygotskian perspective (e.g. Vygotsky, 1978). In this view, the child’s early repertoire of utterances derive their communicative function from evoking the full, rich context of their original use. So, for example, "ball", when used by the 1 yr old, does not refer to just one particular (or class of) round object(s), but to the whole set of social and physical activities and surroundings that typically co-occur in the child’s experience of that object. From this perspective, the utterance "ball" is just one of those activities, one that can be reproduced by the child even when some or most of the other aspects of the relevant situation are absent. It is by comparing such utterances with their original form that particular modifications become evident.

One modification involves a "flattening" of affect, where an utterance that was used with some emotional inflection in its original context, is now repeated without that emotional content. This can occur immediately after its affective use, while much of the originally evoking context is still in effect, or much later, when few, if any, elements, are present. In either case, the flattening of affect can, in itself, be seen as "detaching" the utterance, to some degree, in the sense of reducing the physical properties (e.g. inflection, loudness) it has in common with the original context. This detachment – both temporal and affective – also weakens associations with the reinforcement contingencies of the original context. That is, it fosters a separation, a de-contextualization, from the immediate consequences of that situation. This, in turn can facilitate, among other things, additional self control, by somewhat reducing the motivation for immediate reaction and providing time for a more studied response.

The other evidence of affect control at this early stage involves using a particular utterance – generally one of the common phonemes well-practiced during babbling, such as "da" or "ba" – in a variety of contexts, varying its inflection while holding its phonemic content constant. Different inflections tend to evoke different responses from more sophisticated interlocutors, and this provides the child with a basis for developing their strategic use. Learned patterns – such as, in English, a rising frequency sweep for the interrogative, a rise to a plateau and then fall for concurrence, etc. – can then also be applied to an array of phonemes. That is, the reproduction of such words with the context-appropriate inflection, in the absence of the original model, show that both their phonemic and affective content are eventually subject to the infant’s control.

Both of these processes – the flattening of inflection and its context-dependent manipulation – demonstrate the child’s burgeoning ability to control its affective expression. Both also play a role, I would argue, in the development of symbolic reference. That is, these affect control mechanisms help the child 1) to produce utterances that are disconnected from the particulars of the context in which they were learned, while still retaining some capacity to refer to (evoke) that context and 2) to apply them appropriately across a variety of contexts. Both are critical to an utterance’s function as a symbol.

Given the very early stage of language development at which this affect control appears, and how much of subsequent language development (e.g. the growth of the lexicon, the combinatorics of syntax, the conceptual sophistication of narrative and metaphor) depend upon it, it would seem reasonable to expect that this capacity would likewise have arisen early in the evolution of language. This being the case, we might further expect that an assessment of communication in our nearest primate relatives could reveal rudiments of this fundamental ability and hints about its early adaptive functions.

The study of control in communication – and especially of the suppression of affect – in nonhuman primates presents a vexing challenge to the ethologist, however, because of the inherent ambiguity of a failure to respond. That is, an apparently passive animal may be repressing its response or it may be genuinely unmoved. Given no direct access to the mental states responsible for such behavior, the ethologist must rely on other behavioral cues to justify interpreting observations as cases of affect control versus indifference. Such cues include abrupt, radical changes in behavior, especially those linked to the attentional behavior of others, as well as apparent insusceptibility to emotional contagion, when nearby individuals are engaged in affective displays. In my research on social communication in bonobos at the San Diego Wild Animal Park (Pan paniscus) examples of such interactions have been captured on videotape, and will be presented here for consideration as evidence of a limited degree of control of affective expression in these animals.

The first example involves "charging displays" most often given (especially in our group) by the adult male. These dramatic displays involve pushing or dragging a branch or other large object directly past other individuals at high speed. It is a high arousal display, marked by pilo-erection, facial grimacing, and sometimes loud vocalizations. In response, younger animals give way or rush to their mothers, and at least one adult female in our group, who has a history of antagonism with the male, regularly flees with a fearful expression and sometimes screams as well. Of particular interest here, however, are the reactions of the other two adult females in the group, who typically become completely immobile and show no change in expression for the duration of the display. While the possibility – even likelihood – remains that they are genuinely less intimidated by such displays, which only rarely result in their actually being attacked, it may also be that their lack of overt response helps preclude such attacks. (Note that the female who flees is often chased.) Whatever the cause-and-effect relationship here, the "no-response" strategy has apparently been learned by these animals, the younger of whom was observed to flee such displays when she too was immature. The development of this behavior would seem to involve some capacity for the control of affective expression.

On a few occasions, the male in this group was recorded calmly walking to a distant site to pick up an object (a collapsed cardboard box) which he sometimes used in such displays. Only after carrying it to a place where he had a direct charging line to the other animals (in the dry moat that surrounds their island) did he begin his display and exhibit the pilo-erection, grimacing and vocalizations typically associated with that behavior. While it might be argued that these affective expressions were evoked by the combination of the proximity of the target animals and the armament of the displayer, his calm demeanor during the (clearly anticipatory) preceding activity is similar enough to the above described case of the human infant’s flattened affect, in a somewhat removed context, to warrant comparison.

In another segment, an adult female was videotaped engaged in a (for her) highly atypical bout of somersaulting, tickling, and giggling with an adolescent female. The adult male noticed them from afar and gamboled toward them with a distinct "playface" and the wagging arms of an animal inciting play. However, as soon as he approached, the older female immediately sat up and assumed a demur, adult-like demeanor. This sudden change also brought both the other animals’ play to an abrupt, although temporary, halt. Regardless of her motivation for, or conception of, this act, the female’s behavior clearly demonstrated her capacity to exercise sudden control over emotional expression.

In a final segment, another of our bonobos was recorded making an aggressive but unsuccessful grab for some browse (food) held by another individual. She then sat very still, with a calm facial expression, directing her gaze away, until the second animal finally stopped monitoring her and looked away. At that point the first animal immediately turned and made another grab for the branch, this time successfully. (For other similar examples, see Goodall 1986; de Waal 1986; Whiten & Byrne 1988.)

Such behaviors in nonhuman primates are sometimes described anthropomorphically as "feigning indifference", "premeditation", "saving face" etc. Such characterizations are relevant to my topic in that I believe that the self control abilities that underlie such strategies in HUMANS are directly related to the control of affect in early language development. However, such attributions to the apes are not necessary to an evolutionary argument for the origins of these abilities. It is both easier, and more appropriate, to argue for the adaptive significance of functional behavioral traits than of the invisible "mental states" that may, or may not, underlie them (see Johnson, in press). Regardless of their construal by the animals, such interactions can provide insights into possible contexts where selective pressures for improved self control may have operated in our own prehistory.

It seems reasonable to suggest that, in the above described examples, the animals gained some advantage from so modifying their behavior. This can generally be seen in the behavior’s impact on the response of others, whether in the inhibition of attack, avoiding dispersal from a premature warning that would decrease the effectiveness of a display, or reducing vigilance toward a contested (food) resource. The case of the adult female who abruptly stopped playing when the male showed up is perhaps somewhat more mysterious, but still suggests that "audience effects" were at issue.

Bonobos (like chimpanzees and, most likely, early hominids) live in fission/fusion societies in the wild, where subgroup membership changes frequently (see Kano, 1992). Such a social structure can offer selective advantages to those who can modify their behavior in such a way as to strategically vary the information they make available to their current audience. The control of affective communication would be particularly useful here, since affect-laden signals tend to be relatively conspicuous and are probably the ones most frequently used by others to predict subsequent behavior. To the extent that it pays off to effect the attention and predictions of others, and inasmuch as the ability to do so is mediated by some genetically influenced neuronal structure that enables the context-dependent control of affective expression, differences in these abilities could be translated into the differential reproduction that is critical to evolutionary change. A greater dependence on food sharing or other divisions of labor, such as have often been suggested for early hominids, might create just the sort of increased social stakes required to ratchet these abilities beyond those of the common primate ancestor.

Finally, in looking to the neuronal structures that may mediate such behavior, the most likely candidates probably lie in the prefrontal cortex. In general, the prefrontal cortex is considerably enlarged in humans relative to other primates and is implicated both in the inhibition of (for example, culturally inappropriate) behavior as well as in various planned activities, including language production. Portions of this cortical lobe (for example, what has been mapped, in the human brain, as "Area 13") are apparently involved in the direct inhibition, or modification, of lower brain centers (such as the amygdala and associated structures) of the limbic system, which play a major role in generating emotional expression. Recent comparative research has shown that this area of prefrontal cortex is generally less well developed in apes than in humans. Interestingly enough, however, bonobos DO show some additional development in this area – in terms of greater cell number and columnar differentiation – compared to the other apes (Semendeferri, et al. 1998). Reported behavioral differences in bonobos, compared to chimpanzees, in terms of the former’s reduced tendency toward exaggerated affective responses (Savage-Rumbaugh 1984; de Waal 1989) is particularly interesting in light of these brain data.

Nonetheless, the above-described bonobo behaviors are relatively rare and clearly rudimentary relative to behavior that is common in even infant humans. Thus, both the brain and the behavior data confirm that our species has made significant strides in our evolution of affect control. Before any analysis of the more complex language abilities in humans can proceed, we perhaps need to acknowledge, and account for the emergence of, the fundamental role played by the control of affective expression.

 Conference site: http://www.infres.enst.fr/confs/evolang/