CHAPTER 8

The Evolutionary of the and Their Relationship to Internal Regulatory Variables

JOHN TOOBY and

Evolutionary psychology is an attempt to unify The discovery of a correct - the psychological, social, and behavioral sci- processing description of a psychological ences theoretically and empirically within a is the fundamental clarifying scien- single, mutually consistent, seamless scientific tific step, because each mechanism came into framework. The core of this enterprise is the in- existence and was organized by natural selec- tegration of principles and findings drawn tion in order to carry out its particular of from evolutionary , cognitive , information-processing functions. It is not a , , and but a that the is a with psychology in order to produce high- computer—a physical system that came into resolution maps of . By “human existence to carry out computations. The com- nature,” evolutionary mean the putations were needed to solve the adaptive evolved, reliably developing, -typical problem of regulating behavior successfully. computational architecture of the human Hence the brain (and its subsystems) evolved to , together with the physical structures and carry out specific varieties of computation in processes (in the brain, in development, and in order to regulate behavior so that it was bio- ) that give rise to this information- logically successful—that is, to assemble the in- processing architecture. For evolutionary psy- dividual somatically and neurally, to prevent chologists, all forms of knowledge about prereproductive , to increase the proba- and behavior are relevant, but the pivotal step bility of achieving conditions (social and physi- is using these facts to form accurate models of cal) that would have led to successful reproduc- the information-processing structure of psy- tion in the ancestral world, to reproduce chological mechanisms. successfully, and to assist genetic relatives (in-

114 8. The of the Emotions 115 cluding children) to achieve and maintain con- not limited to one area of human , such as ditions for their own successful . sex, , or acquisition. Instead, In short, the functional subcomponents (pro- it is a dimension cross-cutting all areas of grams) that constitute our psychological archi- human life, as weighted by the strange, tecture were designed by to nonintuitive metric of their cross-generational solve adaptive problems faced by our hunter– statistical effects on direct and kin reproduc- gatherer ancestors by regulating behavior in tion. ways that increased genetic propagation— By “computation,” evolutionary psycholo- what call “.” Against the oth- gists simply mean the organized causation of erwise disordering forces of entropy that per- patterned information input–output relations. vade all of physical reality, natural selection is Natural selection poses adaptive problems of the only process that introduces functional or- behavior regulation, and the mechanisms of the ganization into the of brain evolved to engineer solutions in the form (Tooby, Cosmides, & Barrett, 2003). So, to the of these regulatory input–output relations. Of extent that there is functional organization in course, these computational relations must be the human psychological architecture, it was embodied physically in neural , and must created by, reflects, and is explained by the be designed to develop reliably. operation of natural selection among our an- are not just the products of the , but are cestors. This is why evolutionary psychology is the products of the coordinated interaction of a not a specific subfield of psychology, such as stable genetic inheritance and the - the study of vision, reasoning, or social behav- arily long-enduring features of the environ- ior. It is a way of approaching the science of ment. psychology that produces (or is intended to A model of an evolved neurocomputational produce) stable functional descriptions of the mechanism or program would answer ques- elements of the mind. (Detailed arguments for tions such as these: What information does the these positions can be found in Tooby & program take as input? How is this infor- Cosmides, 1990a, 1990b, 1992a, 2005, and in mation encoded, formatted, and represented Cosmides & Tooby, 1987, 1992, 1997.) as data structures? What operations are per- Researchers less familiar with evolutionary formed on these data structures to transform psychology often equate adaptive problems ex- them into new representations or regulatory el- clusively with short-run threats to physical sur- ements? And how do these procedures and vival. However, survival per se is not central to data structures interact to generate and regu- evolution: All organisms die sooner late behavior? In short, how does each pro- or later. In contrast, genes—which can be gram work in cause-and-effect terms? of as particles of —are poten- tially immortal, and design features spread by promoting the reproduction of the genes that AN EVOLUTIONARY- participate in building them. Survival is signifi- PSYCHOLOGICAL APPROACH cant only insofar as it promotes the reproduc- TO THE EMOTIONS tion of design features into subsequent genera- tions. Survival is no more significant than Although an evolutionary-psychological ap- anything else that promotes reproduction, and proach can be applied to any topic in psychol- is often advantageously risked or sacrificed in ogy, it is especially illuminating when applied the process of promoting reproduction in , to the emotions. To the extent that there is children, or other relatives. Nearly every kind functional order to be found in the mechanisms of event or condition has the potential to have responsible for the emotions, it was forged over some impact on the prospect of reproduction evolutionary time by natural selection acting for , their children, and their rela- on our ancestors. The analysis of adaptive tives. Consequently, selection on neural designs problems that arose ancestrally has led evolu- for functional behavior reaches out to encom- tionary psychologists to apply the concepts and pass, in a network of cause and effect linkages, methods of the evolutionary to scores virtually all of human life, from the subtleties of topics that are relevant to the study of emo- of to attributions of responsi- tion. These include , cooperation, sexual bility to the intrinsic rewards of . The realm attraction, , , parental , of adaptive information-processing problems is , romantic love, the of 116 I. INTERDISCIPLINARY FOUNDATIONS landscape preferences, coalitional aggression, exploration new empirical and theoretical pos- avoidance, , predator avoidance, sibilities obscured by other frameworks. , and relations (for reviews, see Barkow, Cosmides, & Tooby, 1992; Buss, 2005; Crawford & Krebs, 1998; Daly & Wil- AN EVOLUTIONARY- son, 1988; Pinker, 1997). PSYCHOLOGICAL THEORY Indeed, a rich theory of the emotions natu- OF THE EMOTIONS rally emerges out of the core principles of evo- lutionary psychology (Tooby, 1985; Tooby & Both deductions from theoretical evolutionary Cosmides, 1990a; Cosmides & Tooby, 2000; psychology and a large supporting body of em- see also Nesse, 1991). In this chapter, we (1) pirical findings in psychology, biology, and briefly state what we think emotions are and neuroscience support the view that the human what adaptive problem they were designed to mental architecture is crowded with evolved, solve; (2) explain the evolutionary and compu- functionally specialized programs. Each is tai- tational principles that led us to this view; (3) lored to solve a different adaptive problem that identify how the emotions relate to motiva- arose during human evolutionary (or tional and other underlying regulatory vari- before), such as recognition, , ables the is designed to generate , rate regulation, sleep - and access; and (4) using this background, ex- agement, or predator vigilance, and each is ac- plicate in a more detailed way the design of tivated by a different set of cues from the envi- programs and the states they create. ronment. It may strike some as odd to speak about But the existence of all these diverse programs love, jealousy, or disgust in computational itself creates an adaptive problem: Programs terms. “Computation” has an affectless, fla- that are individually designed to solve specific vorless connotation. But if the brain evolved as adaptive problems could, if simultaneously acti- a system of information-processing relations, vated, deliver outputs that conflict with one an- then emotions are, in an evolutionary , other, interfering with or nullifying each other’s best understood as information-processing functional products. For example, sleep and relations—that is, programs—with naturally from a predator require mutually inconsis- selected functions. Initially, the commitment to tent actions, computations, and physiological exploring the underlying computational archi- states. It is difficult to sleep when your heart and tecture of the emotions may seem infelicitous, mind are racing with ,and this is no accident: but viewing them as programs leads to a large Disastrous consequences would ensue if number of scientific payoffs. In particular, the proprioceptive cues were activating sleep pro- claim that emotion is computational does not grams at the same time that the sight of a stalking mean that an evolutionary-psychological ap- lion was activating ones designed for predator proach misconstrues human as evasion. To avoid such consequences, the mind bloodless, affectless, disembodied ratiocina- must be equipped with superordinate programs tion. It is simply the claim that one can describe that override and deactivate some programs the underlying set of informational relation- when others are activated (e.g., a program that ships that explain emotional phenomena, in- deactivates sleep programs when predator eva- cluding the nature of emotional experience. Ev- sion subroutines are activated). Reciprocally, ery mechanism in the brain—whether it does many adaptive problems are best solved by the something categorizable as “cold ” coordinated activation of a specific subset of (such as reasoning, inducing a rule of grammar, programs, with each program being entrained or judging a ) or as “hot cognition” into the computational settings most appropri- (such as computing the intensity of parental ate for the particular adaptive problem being fear, the imperative to strike an adversary, or an faced. For example, predator avoidance may re- escalation in )—depends on an un- quire simultaneous shifts in both and derlying computational organization to give its auditory acuity (see below). To do this, a special operation its patterned structure, as well as a type of program is required that manages and set of neural circuits to implement it physically. harmonizes other programs, aligning each of In these terms, an evolutionary and comp- them into the proper configuration at the right utational view of emotion can open up for time. 8. The Evolutionary Psychology of the Emotions 117

In general, to behave functionally according execute its own ancestral functional logic even to evolutionary standards, the mind’s many under modern conditions. subprograms need to be orchestrated so that How did emotions arise and assume their their joint product at any given time is coordi- distinctive structures? Fighting, , nated to deal with the adaptive challenge being escaping predators, confronting sexual infidel- faced, rather than operating in a self-defeating, ity, experiencing a failure-driven loss in status, discoordinated, and cacophonous fashion. We responding to the death of a family member, argue that such coordination is accomplished and so on each involved conditions, contingen- by a special class of programs: the emotions cies, situations, or event types that recurred in- that evolved to solve these superordinate de- numerable times in evolutionary his- mands. In this view, the best way to understand tory. Repeated encounters with each kind of what the emotions are, what they do, and how situation selected for adaptations that guided they operate is to recognize that mechanism or- information processing, behavior, and the body chestration is the that defines the adaptively through the clusters of conditions, emotions, and explains in detail their design demands, and contingencies characterizing that features. They are the neurocomputational ad- particular class of situation. aptations that have evolved in response to the The payoffs accruing to alternative mutant adaptive problem of matching arrays of mecha- designs for program activation, in interaction nism activation to the specific adaptive de- with recurrent classes of situations, engineered mands imposed by alternative situations programs each of which jointly mobilizes a (Tooby & Cosmides, 1990a; Tooby, 1985; subset of the psychological architecture’s other Cosmides & Tooby, 2000; Nesse, 1991). programs in a particular configuration. Each Thus each emotion evolved to deal with a configuration was selected to deploy computa- particular, evolutionarily recurrent situation tional and physiological mechanisms in a way type. The design features of the emotion pro- that, when averaged over individuals and gen- gram, when the emotion is activated, presume erations, would have led to the most fitness- the presence of an ancestrally structured situa- promoting subsequent lifetime outcome, given tion type (regardless of the actual structure of that ancestral situation type. Thus an emotion the modern world). Hence the exploration of is a bet placed under conditions of uncertainty: the statistical structure of ancestral situations It is the evolved mind’s bet about what internal and their relationship to the mind’s battery of deployment is likely to lead to the best average functionally specialized programs is central to long-term set of payoffs, given the structure mapping the emotions. This is because the and statistical contingencies present in the an- most useful (or least harmful) deployment of cestral world when a particular situation was programs at any given time will depend criti- encountered. Running away in terror, vomiting cally on the exact nature of the situation being in disgust, or attacking in are bets that are encountered. The abstract, distilled, recurrent placed because these responses had the highest characteristics of the situation are reflected in average payoffs for our ancestors, given the the architecture of the emotion. For example, eliciting conditions. because sexual rivals could be advantageously This coordinated adjustment and entrain- driven off by violence or its threat in a substan- ment of mechanisms constitutes a mode of tial fraction of the trillions of ancestral cases of operation for the entire psychological architec- mate , is engineered ture, and serves as the basis for a precise com- to prepare the body physiologically for com- putational and functional definition of each bat, and (when the rival is weak or unwary) emotion state (Tooby & Cosmides, 1990a; motivates the individual to behave violently. In Tooby, 1985; Cosmides & Tooby, 2000). Each modern situations of potential or actual infidel- emotion entrains various other adaptive ity, police and prisons create additional conse- programs—deactivating some, activating oth- quences, and so violence against a sexual rival ers, and adjusting the modifiable parameters of is likely to lead to maladaptive outcomes now. still others—so that the whole system operates However, the design features of jealousy were in a particularly harmonious and efficacious designed to mesh with the long-enduring struc- way when the individual is confronting certain ture of the ancestral world, and not the modern kinds of triggering conditions or situations. world—so the emotion program continues to The conditions or situations relevant to the 118 I. INTERDISCIPLINARY FOUNDATIONS emotions are those that (1) recurred ances- FEAR AS A MODE OF OPERATION trally; (2) could not be negotiated successfully unless there was a superordinate level of pro- Consider the following example. The ances- gram coordination (i.e., circumstances in trally recurrent situation is being alone at which the independent operation of programs night, and a situation detector circuit perceives caused no conflicts would not have selected for cues that indicate the possible presence of a hu- an emotion program, and would lead to emo- man or animal predator. The emotion mode is tionally neutral states of mind); (3) had a rich a fear of being stalked. (In this conceptualiza- and reliable repeated structure; (4) had recog- tion of emotion, there might be several distinct nizable cues signaling their presence; and (5) emotion modes that are lumped together under would have resulted in large fitness costs if an the folk category “fear,” but that are at least had occurred (Tooby & Cosmides, partially distinguishable, computationally and 1990a; Tooby, 1985; Cosmides & Tooby, empirically, by the overlapping but nonidenti- 2000). When a condition or situation of an cal constellation of programs each entrains.) evolutionarily recognizable kind is detected, a When the situation detector signals that one signal is sent out from the emotion program has entered the situation of “possible stalking that activates the specific constellation of and ambush,” the following kinds of mental subprograms appropriate to solving the type of programs are entrained or modified: adaptive problems that were regularly embed- ded in that situation, and deactivates programs 1. There are shifts in and atten- whose operation might interfere with solving tion. You may suddenly hear with far greater those types of adaptive problems. Programs di- clarity sounds that bear on the hypothesis that rected to remain active may be cued to enter you are being stalked, but that ordinarily you subroutines that are specific to that emotion would not perceive or attend to, such as creaks mode, and that were tailored by natural selec- or rustling. Are the creaks footsteps? Is the rus- tion to solve the problems inherent in the trig- tling caused by something moving stealthily gering situation with special efficiency. (Where through the bushes? Signal detection thresh- there was no repeated structure, or there were olds shift: Less evidence is required before you no cues to signal the presence of a repeated respond as if there were a threat, and more true structure, then selection could not build an ad- positives will be perceived at the cost of a aptation to address the situation.) higher rate of false alarms. According to this theoretical framework, an 2. and motivational weightings emotion is a superordinate program whose change. Safety becomes a far higher priority. function is to direct the activities and interac- Other goals and the computational systems tions of the subprograms governing perception; that subserve them are deactivated: You are no ; inference; ; ; longer hungry; you cease to think about how to choice; motivational priorities; categorization charm a potential mate; practicing a new skill and conceptual frameworks; physiological re- no longer seems rewarding. Your planning fo- actions (such as heart rate, endocrine function, cus narrows to the present; worries about yes- immune function, release); reflexes; terday and tomorrow temporarily vanish. behavioral decision rules; motor systems; com- , thirst, and are suppressed. munication processes; level and effort 3. Information-gathering programs are redi- allocation; affective coloration of events and rected: Where is my ? Where are others stimuli; recalibration of probability estimates, who can protect me? Is there somewhere I can situation assessments, values, and regulatory go where I can see and hear what is going on variables (e.g., self-esteem, estimations of rela- better? tive formidability, relative of alternative 4. Conceptual frames shift, with the auto- goal states, efficacy discount rate); and so on. matic imposition of categories such as “danger- An emotion is not reducible to any one cate- ous” or “safe.” Walking a familiar and usually gory of effects, such as effects on , comfortable route may now be mentally tagged behavioral inclinations, cognitive appraisals, or as “dangerous.” Odd places that you normally states, because it involves evolved in- would not occupy—a hallway closet, the structions for all of them together, as well as branches of a tree—suddenly may become sa- other mechanisms distributed throughout the lient as instances of the category “safe” or human mental and physical architecture. “hiding place.” 8. The Evolutionary Psychology of the Emotions 119

5. Memory processes are directed to new re- immobility (the last of these is a common re- trieval tasks: Where was that tree I climbed be- sponse to actual attacks, both in other animals fore? Did my adversary and his friend look at and in humans1). Some of these responses may me furtively the last time I saw them? be experienced as automatic or involuntary. 6. processes change. De- pending on the circumstances, decision rules From the point of view of avoiding danger, may cause you to emit an alarm cry, or be para- these computational changes are crucial: They lyzed and unable to speak. Your face may auto- are what allowed the adaptive problem to be matically assume a species-typical fear expres- solved with high probability, on average over sion. evolutionary time. Of course, in any single case 7. Specialized inference systems are acti- they may fail, because they are only the vated. Information about a lion’s trajectory or evolutionarily computed best bet, based on an- eye direction may be fed into systems for infer- cestrally summed outcomes; they are not a sure ring whether the lion saw you. If the inference bet, based on an unattainable perfect knowl- is yes, then a program automatically infers that edge of the present. the lion knows where you are; if no, then the Whether individuals report consciously ex- lion does not know where you are (the “seeing- periencing fear is a separate question from is-knowing” circuit identified by Baron-Cohen, whether their mechanisms assumed the charac- 1995, as impaired in persons with ). teristic configuration that, according to this This variable may automatically govern theoretical approach, defines the fear emotion whether you freeze in terror or bolt. Are there state. Individuals often behave as if they are in cues in the lion’s behavior that indicate whether the grip of an emotion, while denying they are it has eaten recently, and so is unlikely to be feeling that emotion. We think it is perfectly predatory in the near future? (Savanna- possible that individuals sometimes remain un- dwelling ungulates, such as zebras and wilde- aware of (or lose conscious access to) their beests, commonly make this kind of judgment; emotion states, which is one we do not Marks, 1987.) use subjective experience as the sine qua non of 8. Specialized learning systems are acti- emotion. At present, both the function of con- vated, as the large on fear condition- scious awareness, and the principles that regu- ing indicates (e.g., LeDoux, 1995; Mineka & late conscious access to emotion states and Cook, 1993; Pitman & Orr, 1995). If the threat other mental programs, are and unre- is real, and the ambush occurs, you may experi- solved questions (but see Tooby, Cosmides, ence an -mediated recalibration (as in Sell, Lieberman, & Sznycer, in press). Mapping posttraumatic disorder) that can last for the design features of emotion programs can the remainder of your life (Pitman & Orr, proceed independently of their resolution, at 1995). least for the present. 9. Physiology changes. Gastric mucosa turn white as blood leaves the digestive tract (an- other concomitant of motivational priorities ADAPTATIONIST FOUNDATIONS changing from feeding to emergency motor ac- Adaptations, By-Products, and Noise tivity in pursuit of safety); adrenalin spikes; heart rate may go up or down (depending on Because of the different roles played by chance whether the situation calls for flight or immo- and selection, the evolutionary process builds bility), blood rushes to the periphery, and so on three different types of outcomes into - (Cannon, 1929; Tomaka, Blascovich, Kibler, & isms: (1) adaptations—that is, functional ma- Ernst, 1997); instructions to the musculature chinery built by selection, and usually species- (face and elsewhere) are sent (Ekman, 1982). typical (see Tooby & Cosmides, 1990b, for de- Indeed, the nature of the physiological re- tails and exceptions); (2) by-products of adap- sponse can depend in detailed ways on the na- tations, which are present in the design of or- ture of the threat and the best response option ganisms because they are causally coupled to (see, e.g., Marks, 1987). traits that were selected for (usually species- 10. Behavioral decision rules are activated: typical); and (3) random noise, injected by mu- Depending on the nature of the potential tation and other random processes (often not threat, different courses of action will be poten- species-typical) (Tooby & Cosmides, 1990a, tiated: hiding, flight, self-defense, or even tonic 1990b, 1992a; Williams, 1966). The emotion 120 I. INTERDISCIPLINARY FOUNDATIONS of sexual jealousy is an (Daly, Wil- only that their functional logic is likely to be son, & Weghorst, 1982; Buss, 1994); stress- sophisticated and well engineered to solve an- induced physical deterioration is arguably a by- cestral adaptive problems. product of the flight–fight system; and herita- ble variation in emotional function- The Environment ing (e.g., extreme , morbid jealousy, bi- of Evolutionary Adaptedness polar disorder) is probably noise (Tooby & Cosmides, 1990b). Evidence of the presence (or Behavior in the present is generated by evolved absence) of high degrees of coordination be- information-processing mechanisms that were tween adaptive problems and the design fea- constructed in the past. They were constructed tures of putative adaptations allows research- in the past because they solved adaptive prob- ers to distinguish adaptations, by-products, lems that were recurrently present in the ances- and noise from one another (Williams, 1966; tral environments in which the human line Cosmides & Tooby, 1997). evolved. For this reason, evolutionary psychol- The emotions are often thought of as crude, ogy is both environment-oriented and past- but we expect emotions to be very well- oriented in its functionalist orientation. Adap- designed computational adaptations. Biologists tations become increasingly effective as selec- have found that selection has routinely pro- tion makes their design features more and more duced exquisitely engineered biological ma- complementary to the long-enduring structure chines of the highest order at all scales, from of the world. The articulated features of the ad- genetic error correction and quality control in aptation are designed to mesh with the features assembly to photosynthetic pigments, of the environment that were stable during the the , efficient foraging algo- adaptation’s evolution, so that their interaction rithms, echolocation, and color constancy sys- produced functional outcomes. The regulation tems. Indeed, the best-studied psychological of breathing assumes the presence of certain adaptation—the eye and visual system—has long-enduring properties of the atmosphere been held up for centuries as the apotheosis of and the respiratory system. Vision assumes the engineering excellence, as yet unrivaled by any presence of certain evolutionarily stable prop- human engineer. There is no principled reason erties of surfaces, objects, and terrestrial spec- to expect other neurocomputational (i.e., psy- tral distributions. The digestive lactase chological) adaptations to be less well engi- presupposes an infant diet of milk with lactose. neered than the eye. Although Stephen Jay Fear presupposes dangers in the environment, Gould (1997) and his followers have energeti- and even presupposes higher of cally argued in the popular science literature specific kinds of dangers, given certain cues: that natural selection is a weak evolutionary darkness, spiders, snakes, heights, predators, force, evolutionary biologists, familiar with the open spaces, and so on (Marks, 1987). That is, primary literature, have found it difficult to each emotion program presupposes that cer- take these arguments seriously (Tooby & tain cues signal the presence of a structure of Cosmides, 1997). So although adaptations are events and conditions that held true during the in some abstract sense undoubtedly far from evolution of that emotion. Disgust circuits pre- optimal (and there is genetic noise in all sys- sume a world in which rotten smells signal tox- tems), the empirical evidence falsifies the claim ins or microbial contamination, for example. that evolved computational adaptations tend Accordingly, to understand an adaptation as to be crude or primitive in design, and instead a problem solver, one needs to model the en- supports the opposite view: that our mental during properties of the task environment that machinery—including the emotions—islikely constituted the problem and provided materi- to be very well designed to carry out evolved als that could be exploited for its solution: the functions. For emotion researchers, this means “environment of evolutionary adaptedness,” that their working hypotheses (which are al- or (EEA). Although the human line is thought ways open to empirical revision) should begin to have first differentiated itself from the chim- with the of high levels of evolu- panzee lineage on the African savannahs and tionary functionality, and their research meth- woodlands, the EEA is not a place or time. It is ods should be sensitive enough to detect such the statistical composite of selection pressures organization. This does not mean that emo- that caused the genes underlying the design of tions are well designed for the modern world— an adaptation to increase in frequency until 8. The Evolutionary Psychology of the Emotions 121 they became species-typical or stably persistent als the mind has tagged as (Lieberman (Tooby & Cosmides, 1990a). Thus statistical et al., 2007). regularities define the EEA for any given adap- tation. The conditions that characterize the The Functional Structure EEA are usefully decomposed into a constella- of an Emotion Program Evolved to Match tion of specific environmental regularities that the Evolutionarily Summed Structure had a systematic (though not necessarily un- of Its Target Situation varying) impact on reproduction, and that endured long enough to work evolutionary Each emotion program was constructed by a change on the design of an adaptation. Some of selective regimen consisting of repeated en- these regularities are extremely simple: Dis- counters with a particular kind of evolution- tance from a predator is protection from the arily recurrent situation. By an “evolutionarily predator. Sex with an opposite-sex is recurrent situation,” we mean a cluster of re- more likely to produce offspring than sex with peated probabilistic relationships among a child or a nonhuman. These regularities can events, conditions, actions, and choice conse- equally well include complex conditionals (e.g., quences that endured over a sufficient stretch if one is a male hunter–gatherer and one is hav- of evolutionary time to have favored some vari- ing a sexual liaison with someone else’s mate designs over others. Many of these relation- and that liaison is discovered, then one is the ships were probabilistically associated with target of lethal retributory violence 14% of the cues detectable by , allowing psycho- time). Descriptions of these regularities are es- physical triggers to activate the task- sential parts of the construction of a task analy- appropriate program. sis of the adaptive problem a hypothesized ad- For example, the condition of having a mate aptation evolved to solve (Tooby & Cosmides, plus the condition of the mate’s copulating with 1990a). Conceptualizing the EEA in probabil- someone else constitutes a situation of sexual istic terms is fundamental to the functional def- —a situation that has recurred over inition of emotion that we have presented evolutionary time, even though it has not hap- above and will elucidate below. pened to every individual. Associated with this Each adaptive problem recurred billions or situation were cues reliable enough to allow the trillions of times in the EEA, and so manifested evolution of a “situation detector” (e.g., ob- a statistical and causal structure whose ele- serving a sexual act, flirtation, or even the re- ments were available for specialized exploita- peated simultaneous absence of the suspected tion by design features of the evolving adapta- lovers were cues that could trigger the categori- tion. For example, predators use darkness and zation of a situation as one of infidelity). Even cover to ambush (Marks, 1987). Physical more importantly, there were many necessarily appearance varies with and health or probabilistically associated elements that (Symons, 1979). Among hunter–gatherers, in- tended to be present in the situation of infidel- fants that a primarily cares for are al- ity as encountered among our hunter–gatherer most invariably genetic siblings (Lieberman, ancestors. These additional elements included Tooby, & Cosmides, 2007). Specialized (1) a sexual rival with a capacity for social ac- programs—for predator fear, , tion and violence, as well as allies of the rival; and kin detection, respectively—could evolve (2) a discrete probability that one’s mate had whose configuration of design features embod- conceived with the sexual rival; (3) changes in ied and/or exploited these statistical regulari- the net lifetime reproductive returns of invest- ties, allowing these adaptive problems to be ing further in the relationship; (4) a solved economically, reliably, and effectively. probable decrease in the degree to which the Such specializations, by embodying “innate unfaithful mate’s mechanisms would value the knowledge” about the problem space, operate victim of infidelity (the presence of an alterna- better than any general learning strategy could. tive mate would lower replacement costs); (5) a Children did not have to wait to experience be- cue that the victim of the infidelity was likely to ing ambushed and killed in the dark to pru- have been deceived about a range of past dently modulate their activities. did not events, leading the victim to confront the likeli- need to observe the negative effects of incest, hood that his or her memory was permeated because the human kin detection system mobi- with false information; and (6) a likelihood lizes disgust toward having sex with individu- that the victim’s status and reputation for being 122 I. INTERDISCIPLINARY FOUNDATIONS effective at defending his or her interests in gen- lies and increases the probability that he will be eral would plummet, inviting challenges in challenged by competitors in other domains of other arenas. These are just a few of the many life. The sexual jealousy, anger, and sys- factors that would constitute a list of elements tems have been shaped by the distillation of associated in a probabilistic cluster, and that these (and other) payoff probabilities. Each of would constitute the evolutionary recurrent these recurrent subelements in a situation of structure of a situation of sexual infidelity. The sexual infidelity, and the adaptive circuits they emotion of sexual jealousy evolved in response require, can be added together to form a gen- to these properties of the world, and there eral theory of sexual jealousy, as well as a the- should be evidence of this in its computational ory of the functional coactivation of linked design. programs (such as anger and shame). Emotion programs have evolved to take such Hence the emotion of sexual jealousy consti- elements into account, whether they can be tutes an organized mode of operation specifi- perceived or not. Thus not only do cues of a sit- cally designed to deploy the programs govern- uation trigger an emotion mode, but embedded ing each psychological mechanism, so that each in that emotion mode is a way of seeing the is poised to deal with the exposed infidelity. world and feeling about the world related to Physiological processes are prepared for such the ancestral cluster of associated elements. De- things as violence, , and the pending on the intensity of the jealousy evoked, withdrawal of investment; the goal of deter- less and less evidence will be required for indi- ring, injuring, or murdering the rival emerges; viduals to believe that these conditions apply to the goal of punishing, deterring, or deserting their personal situation. Individuals with mor- the mate appears; the to make oneself bid jealousy, for example, may hallucinate more competitively attractive to alternative counterfactual but evolutionarily thematic con- mates emerges; memory is activated to tents, such as seeing their mates having sex reanalyze the past; confident assessments of the with someone else (Mowat, 1966; Shepherd, past are transformed into ; the general 1961). This leads many to consider emotions estimate of the reliability and trustworthiness “irrational,” but this property was selected for of the opposite sex (or indeed everyone) may because it allows emotional computation to go decline; associated shame programs may be beyond the evidence given, producing correct triggered to search for situations in which the responses (when averaged over evolutionary individual can publicly demonstrate acts of vio- time). lence or that work to counteract To the extent that situations exhibited a an (imagined or real) social perception of structure repeated over evolutionary time, their weakness; and so on. statistical properties would be used as the basis It is the relationship between the summed for natural selection to build an emotion pro- details of the ancestral condition and the de- gram whose detailed design features were tai- tailed structure of the resulting emotion pro- lored for that situation. This would be accom- gram that makes this approach so useful for plished by selection acting over evolutionary emotion researchers. Each functionally distinct time, differentially incorporating program emotion state—fear of predators, , sexual components that dovetailed with individual jealousy, rage, , and so on—will corre- items on the list of properties probabilistically spond to an integrated mode of operation that associated with the situation. functions as a solution designed to take advan- For example, ancestrally a male’s ability to tage of the particular structure of the recurrent inflict costs through violence (his “formid- situation or triggering condition to which that ability”) was associated with his status and emotion corresponds. This approach can be reputation for defending his interests. More- used to create theories of each individual emo- over, the fitness consequences of being cuck- tion, through four steps: (1) Reconstruct the olded are great, and males have become moti- clusters of properties of ancestral situations; (2) vated by design to resist this outcome. If a analyze what behavioral and somatic alter- mate’s mate is sexually unfaithful and this infi- ations would solve the adaptive problem posed delity becomes public, this advertises a weak- by the recurrent situation (or minimize the ness previously unappreciated by those who damage it causes); (3) construct a provisional know him best. This decrease in perceived model of the program architecture of the emo- formidability decreases his value to his male al- tion that could generate the necessary 8. The Evolutionary Psychology of the Emotions 123 mechanism-, body-, and behavior-regulating ambush an enemy, having few friends, experi- outputs, including the cues used, the regulatory encing the death of a , being sick, having variables the emotion needs to track, and so experienced a public , having others act on; and (4) design and conduct experiments in a way that damages you without regard for and other investigations to test each hypothe- your welfare, having injured a valued other sized design feature of the proposed emotion through insufficient consideration of self–other program, revising them as necessary. behavioral tradeoffs, and having a baby. It is also important to understand that evo- 2. The adaptive problem. Identifying the lutionarily recurrent situations can be arrayed adaptive problem means identifying which or- along a spectrum in terms of how rich or skele- ganismic states and behavioral sequences will tal the set of probabilistically associated ele- lead to the best average functional outcome for ments defining the recurrent situation is. Richly the remainder of the lifespan, given the situa- structured situations—such as sexual infidelity, tion or condition. For example, what is the best exposure to potential vectors, or preda- course of action when others take the products tor ambush—will support a richly substruc- of your labor without your ? What is tured emotion program in response to the nu- the best course of action when you are in a de- merous ancestrally correlated features each pleted nutritional state? What is the best course manifests: Many detailed adjustments will be of action when a makes a sexual ap- made to many psychological mechanisms as in- proach? structions for the mode of operation. In con- 3. Cues that signal the presence of the situa- trast, some recurrent situations have less struc- tion. For example, low blood sugar signals a ture (i.e., they share fewer properties), and so depleted nutritional state; the looming ap- the emotion mode makes fewer highly special- proach of a large, fanged animal signals the ized adjustments, imposes fewer specialized presence of a predator; seeing your mate hav- and compelling interpretations and behavioral ing sex with another signals sexual infidelity; inclinations, and so on. For example, surges of finding yourself often alone, rarely the recipi- or are an emotion program that ent of beneficent acts, or actively avoided by evolved to respond to the recurrent situation of others signals that you have few friends. encountering unexpected positive events. The 4. Situation-detecting . A multi- class of events captured by “unexpectedly posi- modular mind must be full of “demons”—al- tive” is extremely broad and general, and such gorithms that detect situations. The New events have only a few additional properties in Hacker’s Dictionary defines a “demon” as a common. Emotion programs at the most gen- “portion of a program that is not invoked ex- eral and skeletal end of this spectrum corre- plicitly, but that lies dormant waiting for some spond to what some call “” (happiness, condition(s) to occur” (Raymond, 1991, , excitement, , playfulness, p. 124). Situation-detecting subprograms lie , etc.). dormant until they are activated by a specific constellation of cues that precipitates the anal- ysis of whether a particular ancestral situation HOW TO CHARACTERIZE has arisen. If the assessment is positive, it sends AN EMOTION the signal that activates the associated emotion program. Emotion demons need two kinds of To characterize an emotion adaptation, one subroutines: must identify the following properties of envi- a. Algorithms that monitor for situation- ronments and of mechanisms. defining cues. These programs include percep- tual mechanisms, proprioceptive mechanisms, 1. An evolutionarily recurrent situation or and situation-representing mechanisms. They condition. A “situation” is a repeated structure take the cues in point 3 above as input. of environmental and organismic properties, b. Algorithms that detect situations. These characterized as a complex statistical compos- programs take the output of the monitoring al- ite of how such properties covaried in the envi- gorithms and targeted memory registers in ronment of evolutionary adaptedness. Exam- point a as input, and through integration, ples of these situations are being in a depleted probabilistic weighting, and other decision cri- nutritional state, competing for maternal atten- teria, identify situations as absent or present tion, being chased by a predator, being about to with some probability and with some index of 124 I. INTERDISCIPLINARY FOUNDATIONS the magnitude of the fitness consequences in- puts. When the representational space assumes herent in the situation. certain configurations, an interpretation is trig- The assignment of a situation interpretation gered that activates the associated emotion to present circumstances involves a problem in program—corresponding approximately to signal (Tooby & Cosmides, what others have called a “” 1990a; Swets, Tanner, & Birdsall, 1961; see (see, e.g., Lazarus & Lazarus, 1994). It is im- also Gigerenzer & Murray, 1987). Animals portant to recognize that the evolutionary past should be designed to detect what situation frames the experienced present, because these they are in on the basis of cues, stored regula- situation-detecting algorithms provide the di- tory variables, and specialized interpretation mensions and core elements out of which many algorithms. Selection will not shape decision cross-culturally recurring representations of rules so that they act solely on the basis of what the world are built. To some extent, the world is most likely to be true, but rather on the basis we inhabit is shaped by the continuous inter- of the weighted consequences of acts, given pretive background commentary provided by that something is held to be true. Should you these mechanisms. walk under a tree that might conceal a preda- 5. Algorithms that assign priorities. A given tor? Even if the algorithms assign a 51% (or world state may correspond to more than one even 98%) probability to the tree’s being situation at a time; for example, you may be leopard-free, under most circumstances an nutritionally depleted and in the presence of a evolutionarily well-engineered decision rule predator. The prioritizing algorithms define should cause you to avoid the tree—to act as if which emotion modes are compatible (e.g., the leopard were in it. The benefits of calories hunger2 and ) and which are mutually saved via a shortcut, scaled by the probability exclusive (e.g., feeding and predator escape). that there is no leopard in the tree, must be Depending on the relative importance of the weighed against the benefits of avoiding be- situations and the reliability of the cues, the coming catfood, scaled by the probability that prioritizing algorithms decide which emotion there is a leopard in the tree. Because the costs modes to activate and deactivate, and to what and benefits of false alarms, misses, hits, and degree. Selection, through ancestral mutant ex- correct rejections are often unequal, the deci- periments, would have sorted emotions based sion rules may still treat as true situations that on the average importance of the consequences are unlikely to be true. In the modern world, stemming from each, and the extent to which this behavior may look “irrational” (as is the joint activation was mutually incompatible or case with many phobias), but we do it because facilitating. (Prioritizing algorithms can be such decision were adaptive under an- thought of as a supervisory system operating cestral conditions, given ancestral payoff asym- over all of the emotions.) metries. That is, they were “ecologically ratio- 6. An internal communication system. nal” (Tooby & Cosmides, 1990a; Haselton & Given that a situation has been detected, the in- Buss, 2003). ternal communication system sends a situation- Situation-detecting algorithms can be of any specific signal to all relevant programs and degree of complexity, from demons that moni- mechanisms; the signal switches them into the tor single cues (e.g., “snake present”) to algo- appropriate adaptive emotion mode. In addi- rithms that carry out more complex cognitive tion, information is fed back into the emotion assessments of situations and conditions program from other programs and systems that (LeDoux, 1995; Lazarus & Lazarus, 1994; assess body states and other regulatory vari- Tooby & Cosmides, 1990a). Inherent in this ables, which may govern the intensity, trajec- approach is the expectation that the human tory, supplantation, or termination of the mind has a series of evolved subsystems emotion. Along with the and moti- designed to represent events in terms of vational systems, the emotions are embedded evolutionarily recurrent situations and situa- in and partly responsible for what might be tional subcomponents. The operations of these called “feeling computation.” In this view, the representational systems are not necessarily richly textured representations we experience consciously accessible. By their structure, they as feeling constitute our conscious access to a impose an evolutionary organization on repre- high-bandwidth system of computational de- sentational spaces that are updated by data in- and program interfaces that amalgamate 8. The Evolutionary Psychology of the Emotions 125 valuation information with other representa- point 7 above). Some such processes are dis- tions to guide decision making and to cussed in this section. recalibrate decisions in an ongoing way (see, e.g., Tooby et al., 2003). Goals Some modes of activation of the psychologi- cal architecture are accompanied by a charac- The cognitive mechanisms that define goal teristic feeling state, a certain quality of experi- states and choose among goals in a planning ence. The fact that we are capable of becoming process should be influenced by emotions. For aware of certain physiological states—our example, vindictiveness—a specialized subcate- thumping, bowels evacuating, stomachs gory of anger—may define “injuring the of- tightening—is surely responsible for some of fending party” as a goal state to be achieved. the evoked by emotion states that en- (Although the evolved functional logic of this train such responses. The fact that we are process is deterrence, this function need not be capable of becoming aware of certain mental represented, either consciously or uncon- states—such as the magnitude of certain regu- sciously, by the mechanisms that generate the latory variables or the retrieved of vindictive behavior.) past events—is probably responsible for other qualia. In our view, the characteristic feeling Motivational Priorities state that accompanies an emotion mode re- sults (in part) from mechanisms that allow us Mechanisms involved in hierarchically rank- to sense the signal activating and deactivating ing goals or calibrating other kinds of moti- the relevant programs, as well as signals com- vational and reward systems should be municating necessary parameters and variable emotion-dependent. What may be extremely magnitudes to the various programs. Such unpleasant in one state, such as harming an- internal sensory mechanisms—analogous to other, may seem satisfying in another state proprioception—can be selected for if there are (e.g., aggressive competition may facilitate mechanisms requiring as input the information counterempathy). Different evolutionarily re- that a particular emotion mode has been acti- current situations predict the presence— vated. (This might be true, for example, of visible or invisible—of different opportunities, mechanisms designed to inhibit certain risks, and payoffs, so motivational thresholds -driven actions when the conditions and valences should be entrained. For exam- are not auspicious.) ple, a loss of face should increase the motiva- 7. Each program and physiological mecha- tion to take advantage of opportunities for nism entrained by an emotion program must status advancement, and should decrease at- have associated algorithms that regulate how it tention to attendant costs. responds to each emotion signal. These algo- rithms determine whether the mechanism Information-Gathering should switch on or switch off, and if on, what emotion-specialized performance it will imple- Because establishing which situation one is in ment. For example, there should be algorithms has enormous consequences for the appropri- in the auditory system that, upon detecting the ateness of behavior, the process of detection fear signal (see point 6), reset signal detection should in fact involve specialized inference pro- thresholds, increasing acuity for predator- cedures and specialized motivations to discover relevant sounds. whether certain suspected facts are true or false. What one is curious about, what one finds interesting, and what one is obsessed with WHAT KINDS OF PROGRAMS discovering should all be emotion-specific. CAN EMOTIONS MOBILIZE? Imposed Conceptual Frameworks Any controllable biological or neurocomputa- tional process that, by shifting its performance Emotions should prompt construals of the in a specifiable way, would lead to enhanced world in terms of concepts that are appropriate average fitness outcomes should have come to to the decisions that must be made. When one be partially governed by emotional state (see is angry, domain-specific concepts such as so- 126 I. INTERDISCIPLINARY FOUNDATIONS cial agency, fault, responsibility, and punish- Physiology ment will be assigned to elements in the situa- tion. When one is hungry, the –nonfood Each organ system, tissue, or process is a po- distinction will seem salient. When one is en- tential candidate for emotion-specific regula- dangered, safety categorization frames will ap- tion, and “” is insufficiently specific to pear. The world will be carved up into catego- capture the detailed coordination involved. ries based partly on what emotional state an Each emotion program should send out a dif- individual is in. ferent pattern of instructions (to the face and limb muscles, the autonomic system, etc.), to the extent that the problems embedded in the Perceptual Mechanisms associated situations differ. This leads to an ex- Perceptual systems may enter emotion-specific pectation that different constellations of effects modes of operation. When one is fearful, acuity will be diagnostic of different emotion states of hearing may increase. Specialized perceptual (Ekman, Levenson, & Friesen, 1983). Changes inference systems may be mobilized as well: If in circulatory, respiratory, and gastrointestinal you’ve heard rustling in the bushes at night, hu- functioning are well known and documented, man and predator figure detection may be par- as are changes in endocrinological function. We ticularly boosted, and not simply visual acuity expect thresholds regulating the contraction of in general. In fact, nonthreat interpretations various muscle groups to change with certain may be depressed, and the same set of shadows emotion states, reflecting the probability that will “look threatening”—that is, given a spe- they will need to be employed. Similarly, im- cific threatening interpretation such as “a man mune allocation and targeting may vary with with a knife”—or not, depending on emotion disgust, with the potential for injury, or with state. the demands of extreme physical exertion.

Memory Communication and Emotional Expressions The ability to call up particularly appropriate Emotion programs are expected to mobilize kinds of information out of long-term memory many emotion-specific effects on the ought to be influenced. A woman who has just subcomponents of the human psychological ar- found strong evidence that her has chitecture relevant to communication. Most been unfaithful may find herself flooded by a notably, many emotion programs produce torrent of memories about small details that characteristic species-typical displays that seemed meaningless at the time but that now fit broadcast to others the emotion state of an in- into an interpretation of covert activity. We dividual (Ekman, 1982). Ekman and his col- also expect that what is stored about present leagues have established in a careful series of experience will also be differentially regulated. landmark studies that many emotional expres- Important or shocking events, for example, sions are human universals, both generated and may be stored in great detail (as has been recognized reliably by humans everywhere they claimed about “flashbulb memories”), but have been tested (Ekman, 1994). Indeed, many other, more moderate emotion-specific effects emotional expressions appear to be designed to may occur as well. be informative, and these have been so reliably informative that humans have coevolved auto- Attention mated interpreters of facial displays of emo- tion, which decode these public displays into The entire structure of attention, from percep- knowledge of others’ mental states. tual systems to the contents of high-level rea- Two things are communicated by an authen- soning processes, should be regulated by emo- tic :3 (1) that the associ- tional state. If you are worried that your spouse ated emotion program has been activated in an is late and might have been injured, it is individual, providing observers with informa- hard to concentrate on other ongoing tasks tion about the state of that individual’s mental (Derryberry & Tucker, 1994), but easy to con- programs and physiology (e.g., “I am afraid”); centrate on danger scenarios. Positive emotions and (2) the identity of the evolutionarily recur- may broaden attentional focus (Fredrickson, rent situation being faced, in the estimation of 1998). the signaler (e.g., the local world holds a dan- 8. The Evolutionary Psychology of the Emotions 127 ger). Both are highly informative, and emo- For this reason, the existence of a distinctive tional expressions provide a continuous com- expression is not a necessary aspect of an emo- mentary on the underlying of things tion, nor should it be part of its definition. Jeal- to companions. This provokes the question: ousy and guilt are both genuine emotions lack- Why did selection build facial, vocal, and pos- ing distinctive signals. tural expressions at all? More puzzlingly, why Precisely because we are designed to monitor are they often experienced as automatic and in- broadcast emotions, our attention goes dispro- voluntary? The apparent selective disadvan- portionately to the subset of emotions that do tages of honestly and automatically broadcast- come equipped with emotional expressions. We ing one’s emotional state have led Fridlund think it likely that this has had an impact on (1994), for example, to argue that expressions the history of emotion research—specifically, must be voluntary and intentional communica- that the emotions associated with distinctive tions largely unconnected to emotion state. But expressions have been unnecessarily considered even when people deliberately lie, microex- “primary” or “fundamental.” pressions of face and voice often leak out Finally, many features of facial expressions (Ekman, 1985), suggesting that certain emo- may not just be arbitrary, but may be reliable tion programs do in fact create involuntarily indicators of an emotion state. Many seem to emitted signals that reliably broadcast the per- be functional concomitants of the activity asso- son’s emotion state and that are difficult to ciated with the emotion (such as eyes widening override. Why? or hyperventilation). Others may be signals First, natural selection has shaped emotion that are nonarbitrary; that is, they remove bar- programs to signal their activation, or not, on riers to the correct assessment of aspects of the an emotion-by-emotion basis. For each emo- that the benefits by dem- tion program considered by itself (jealousy, onstrating. For example, the anger expression , disgust, predatoriness, parental may be designed to maximize the perception of love, sexual attraction, , fear), there strength—an advertisement of a property rele- was a net benefit or cost to having others know vant to the , and not just an arbi- that , averaged across individuals trary signal to others that one is angry (Sell, over evolutionary time. For those recurrent sit- Tooby, & Cosmides, in press-b). That is, there uations in which, on average, it was beneficial may be functional why the anger face to share one’s emotion state (and hence assess- has the characteristics it does, rather than con- ment of the situation) with those one was with, sisting of ear flapping or nose twitching. Simi- species-typical facial and other expressions of larly, the baring of teeth may be combat prepa- emotion were constructed by selection. For ex- ration and advertisement (Archer, 1988); the ample, fear was plausibly beneficial to signal, narrowing of the pupil may be preparation for because it signaled the presence of a danger the detection of fast motion; and so on. that might menace one’s kin and cooperators as well, and it also informed others in a way that Behavior might recruit assistance. Guilt was not selected to cause a presentation with an unambiguous, All psychological mechanisms are involved in distinctive signal. the generation and regulation of behavior, so Nevertheless, averaged over evolutionary obviously behavior will be regulated by emo- time, it was functional for the organism to sig- tion state. More specifically, however, mecha- nal the activation of only some emotion states. nisms proximately involved in the generation The conditions favoring signaling an emotion of actions (as opposed to such processes as face are hard to meet (for conditions and discus- recognition, which are only distally regulatory) sion, see Tooby & Cosmides, 1996b; Cosmides should be very sensitive to emotion state. Not & Tooby, 2000a). Consequently, only some only may highly stereotyped behaviors of cer- emotions out of the total species-typical set are tain kinds be released (as during sexual arousal associated with distinctive, species-typical fa- or rage, or as with species-typical facial expres- cial expressions. There should be a larger set of sions and body ), but more complex emotions that have no automatic display. action generation mechanisms should be regu- Moreover, emotions that lack a display are not lated as well. Specific acts and courses of action necessarily less fundamental or less anchored in will be more available as responses in some the evolved architecture of the human mind. states than in others, and more likely to be im- 128 I. INTERDISCIPLINARY FOUNDATIONS plemented. Emotion mode should govern the one’s coalition had in the construction of organized behavioral sequences (Chagnon, 1983). Hirshleifer (1987) and Frank that solve adaptive problems. (1988) are evolutionary economists who have Biologists, psychologists, and economists pursued this logic the furthest, arguing that who adopt an evolutionary perspective have many social behaviors are the result of such recognized that can be used to “commitment problems.” model many forms of social interactions (Maynard Smith, 1982). If the EEA imposes Specialized Inference certain evolutionarily repeated games, then the “strategies” (the evolved cognitive programs Research in evolutionary psychology has that govern behavior in those contexts) should shown that “thinking” or reasoning is not a evolve in the direction of choices that lead to unitary category, but is carried out by a variety the best expected fitness payoffs. The strategy of specialized mechanisms. So, instead of emo- activated in the individual should match the tion’s activating or depressing “thinking” in game (e.g., exchange) and the state of play in general, the specific emotion program activated the game (e.g., having just been cheated)—a should selectively activate appropriate special- process that requires the system of cues, situa- ized inferential systems, such as cheater detec- tion detection, and so on, already discussed. So tion (Cosmides, 1989; Cosmides & Tooby, different emotion and inference programs or 1989, 1992), bluff detection (Cosmides & subprograms may have evolved to correspond Tooby, 1989), precaution detection (Fiddick, to various evolved games, including zero-sum Cosmides, & Tooby, 2000), attributions of competitive games, positive-sum exchange blame and responsibility, and so on. For exam- games, coalitional lottery games, games of ag- ple, fear could influence precautionary reason- gressive competition corresponding to ing (Boyer & Liénard, 2006), competitive loss “chicken,” and so on (for exchange, see could regulate bluff detection, and so on. Cosmides, 1989; Cosmides & Tooby, 1992). Corresponding emotion programs guide the in- Reflexes dividual into the appropriate interactive strat- egy for the social “game” being played, given Muscular coordination, tendency to blink, the state of play. Surprisingly, for some games, threshold for vomiting, shaking, and many rigid obligatory adherence to a prior strategy other reflexes are expected to be regulated by throughout the game is better than the ability emotion programs to reflect the demands of the to revise and change strategies (“voluntarily”) evolved situation. in the light of events. If an individual contem- plating a course of action detrimental to you Learning knows you will take , regardless of the magnitude of the punishment to you that this Emotion mode is expected to regulate learning might unleash, then that individual will be less mechanisms. What someone learns from stim- likely to take such harmful action. This may uli will be greatly altered by emotion mode, be- translate into emotion programs in which the cause of attentional allocation, , desire to attempt certain actions should be situation-specific inferential algorithms, and a overwhelming, to the point where the actions host of other factors. Emotion mode will cause are experienced as compulsory. In the grip of the present context to be divided up into such programs, competing programs, including situation-specific, functionally appropriate cat- the normal integration of prudential concerns egories so that the same stimuli and the same and social consequences, are muted or termi- environment may be interpreted in radically nated. For example, the desire to avenge a mur- different ways, depending on emotion state. der or an infidelity is often experienced in this For example, which stimuli are considered sim- way, and resulting from this desire are ilar should be different in different emotion even culturally recognized as “crimes of pas- states, distorting the shape of the individual’s sion” (Daly & Wilson, 1988). In modern state psychological “similarity space” (Shepard, , where there are police who are paid 1987). Highly specialized learning mechanisms to punish and otherwise enforce agreements, it may be activated, such as those that control is easy to underestimate the importance that food aversions (Garcia, 1990), predator learn- deterrence based on the actions of oneself and ing (Mineka & Cook, 1993), or fear condition- 8. The Evolutionary Psychology of the Emotions 129 ing (LeDoux, 1995). Happiness is expected to person or condition that was subsequently lost signal the energetic opportunity for play, and to (possibly as a consequence); or (3) gradual rec- allow other exploratory agendas to be ex- ognition by situation detectors that one’s long- pressed (Frederickson, 1998). term pattern of effort and time expenditure has not led to a sufficient level of evolutionarily Mood, Energy Level, Effort Allocation meaningful reward, when implicitly compared and to alternative life paths (the condition of Dick- ens’s Scrooge). Discrepancies between expected Overall metabolic budget will be regulated by and actual payoff can occur in the other direc- emotion programs, as will specific allocations tion as well: Joy, or a precipitated surge of hap- to various processes and facilitation or inhibi- piness, is an emotion program that evolved to tion of specific activities. The effort that it respond to the condition of an unexpectedly takes to perform given tasks will shift accord- outcome. It functions to recalibrate previ- ingly, with things being easier or more effortful, ous value states that led to underinvestment in depending on how appropriate they are to the or underexpectation for the successful activities situation reflected by the emotion (Tooby & or choices. Moreover, energy reserves that were Cosmides, 1990a). Thus fear will make it more being sequestered under one assumption about difficult to attack an antagonist, whereas anger future prospects can be released, given new, will make it easier. The with which more accurate expectations about a more plen- a situation has been identified (i.e., emotional tiful or advantageous future. Similarly, one can clarity) should itself regulate the effortfulness be informed of bad outcomes to choices not of situation-appropriate activities. made: For example, one may find out that a (itself an emotional state) should inhibit the ex- company one almost invested in went bank- penditure of energy on costly behavioral re- rupt, or that the highway one almost took was sponses and should motivate more information snowed in. Information of this kind leads to a gathering and information analysis. Nesse strengthening of the decision variables used (1991) has suggested that the function of mood (experienced as ), which is sometimes is to reflect the propitiousness of the present mistaken for pleasure in the misfortune of oth- environment for action—a hypothesis with ers. Reciprocally, one can be informed of good many merits. We have hypothesized (Tooby & outcomes to choices not made, which will be Cosmides, 1990a) a similar function of mood, experienced as unpleasant. based on recognizing that the action–reward Moreover, the functional definition of emo- ratio of the environment is not a function of the tion given here invites the possibility that many environment alone, but an interaction between well-known mental states should be recognized the structure of the environment and the indi- as emotion states—such as the engen- vidual’s present of it. (By “un- dered by infectious illness, coma, shock, the ap- derstanding,” we mean the correspondence be- preciation of , homesickness, sexual tween the structure of the environment, the arousal, confusion, nausea, and so on. For ex- structure of the algorithms, and the weightings ample, when you are sick, initiating actions and other information they use as parameters.) and going about your daily activities is more The phenomenon that should regulate this as- effortful than usual; your is to stay pect of mood is a perceived discrepancy be- home and lie still. Although you feel as if your tween expected and actual payoff. The suspen- energy reserves are depleted, at a physical level sion of behavioral activity accompanied by the same fat reserves and digestively delivered very intense cognitive activity in depressed peo- glucose are available. Malaise is a computa- ple looks like an effort to reconstruct models of tional state, not a physical one, and is designed the world so that future action can lead to pay- to cope with the adaptive problem of illness: It offs, in part through stripping away previous shunts energy from behavior to the immune valuations that led to unwelcome outcomes. system, and possibly signals the need for aid. Depression should be precipitated by (1) a Similarly, when situation-detecting algorithms heavy investment in a behavioral enterprise detect the presence of a very grave internal in- that was expected to lead to large payoffs that , or the potential for one as indicated by a either failed to materialize or were not large major blow, these may trigger a mode of opera- enough to justify the investment; or (2) insuffi- tion of the psychological architecture that is de- cient investment in maintaining a highly valued signed to prevent any discretionary movement: 130 I. INTERDISCIPLINARY FOUNDATIONS coma. The function of coma, in a world before index up-regulates the weight put on i‘s wel- hospitals, was to prevent further injury from fare, while a low kinship index has little effect being done, minimize blood loss and internal on the to treat i altruistically. This is hemorrhaging, and allow the mobilization of one element that up-regulates the emotion of the body’s resources toward repair of immedi- love, attachment, or caring. Independently, the ate threats to life. Note that a coma is not a kinship index is fed as one of many inputs into physically mandated state of paralysis; it is a the “sexual value estimator.” Its function is to computational state—technically, “a state of compute a magnitude, “sexual value” (SVi), unconsciousness from which the patient cannot which regulates the extent to which the actor is be roused” (Miller, 1976, p. 46), or “un- motivated to value or disvalue sexual contact arousable unresponsiveness” (Berkow, 1992, p. with individual i. As with , many fac- 1398). It can occur even when there has been tors (e.g., health, age, ) sexual no damage to the motor system. value, but a high kinship index renders sexual valuation strongly negative, while a low kin- ship index is expected to have little effect on INTERNAL REGULATORY VARIABLES sexual valuation. The system takes as input AND FEELING COMPUTATION two cues, whose values must themselves be stored and updated as regulatory variables. We expect that the architecture of the human The first is maternal perinatal association (i.e., mind, by design, is full of registers for evolved whether an older sibling observes his or her variables whose function is to store summary mother caring for a younger sibling as an in- magnitudes that are useful for regulating fant), and the second is duration of coresidence behavior and making inferences involving valu- between birth and the end of the period of pa- ation. These are not explicit concepts, repre- rental investment. These two cues are pro- sentations, goal states, beliefs, or , but cessed to set the value of the kinship index for rather indices that acquire their meaning via each familiar childhood companion. This sys- the evolved behavior-controlling and tem was designed by natural selection to detect computation-controlling procedures that ac- which familiar others were close genetic rela- cess them. That is, each has a location embed- tives; to create a magnitude corresponding to ded in the input–output relations of our the degree of genetic relatedness; and then to evolved programs, and their function inheres in deploy this information to motivate both a sex- the role they play in the decision of these ual aversion between brothers and sisters, and the programs. a disposition to behave altruistically toward For example, in our recent mapping of the siblings. architecture of the human kin detection system, An internal regulatory variable like the kin- we have identified a series of regulatory vari- ship index or the welfare tradeoff ratio acquires ables needed to make the system work func- its meaning and functional properties fromits re- tionally and to explain the data (Lieberman et lationship to the programs that compute it, and al., 2007). For example, for each familiar indi- from the downstream decisions or processes that vidual i, the system computes and updates a it regulates. The claim is not that such computa- continuous variable, the “kinship index” (Ki), tions and their embedded variables are deliber- which corresponds to the system’s pairwise es- ate or consciously accessible. We think that they timate of genetic relatedness between self and i. are usually nonconscious or implicit. Outputs of When the kinship index is computed or up- processes that access these variables may be con- dated for a given individual, the magnitude is sciously experienced—as disgust (at the pros- taken as input to procedures that are designed pect of sex with a sibling), for them, to regulate kin-relevant behaviors in a fitness- fear (on their behalf), grief (at their loss), and so promoting way. For the case of altruism, the on. Indeed, we think that it may be possible even- kinship index is fed as one of many inputs to tually to arrive at a precise description of compu- the “welfare tradeoff ratio estimator,” whose tational understructure subserving the world of function is to compute a magnitude, the “in- feeling, by considering feeling to be a special trinsic welfare tradeoff ratio” (intWTRi), which form of computation that evolved to deal with regulates the extent to which the actor is intrin- the world of valuation. sically disposed to trade off his or her own wel- Because the computational mapping of moti- fare against that of individual i. A high kinship vational systems and emotion programs is a 8. The Evolutionary Psychology of the Emotions 131 new enterprise, at present it is difficult to know (Ekman, 1973; Brown, 1991), and its early the full range of internal regulatory variables ontogenetic development (Stenberg, Campos, that our psychological architecture is designed & Emde, 1983; Stenberg & Campos, 1990), it to compute and access. On adaptationist seems likely that anger is an adaptation de- grounds, we suspect that the full set may in- signed by natural selection. If so, then its com- clude a surprising variety of registers for spe- putational structure (i.e., what variables cause cialized magnitudes, corresponding to such anger, what behavioral patterns are enacted by things as these: how valuable a mate is, a child it, and what variables cause it to subside) might is, one’s own life is, and so on; how stable or be usefully illuminated by testing predictions variable the food supply is over the long term; derived by reference to the selection pressures the distribution of condition-independent mor- that designed them. tality in the ; one’s expected future life- Humans evolved embedded in small-scale span or period of efficacy; how good a friend social networks involving both cooperation someone has been to you; the extent of one’s and conflict. In many situations, each individ- social support; the aggressive formidability for ual has open to him or her a range of al- self or others (i.e., the ability to inflict costs); ternative behaviors that embody—as one the sexual value of self and others; one’s status, dimension—a spectrum of possible tradeoffs as well as the status of the coalition one be- between the individual’s own welfare and the longs to; present energy stores; one’s present welfare of one or more others. By choosing one health; the degree to which subsistence requires course of conduct, the individual is intention- action; and so on. However, even fo- ally or unintentionally expressing what can be cusing on one small set of internal regulatory termed a welfare tradeoff ratio with respect to variables, welfare tradeoff ratios, offers to clar- the affected party or parties. For example, an ify the functional architecture of several emo- individual might act in a way that weights the tions, including anger, guilt, and gratitude. welfare of another person slightly or not at all (e.g., being late, theft, marital abandonment, rape, burning down someone’s house for the ANGER AS AN EVOLVED fun of it), in a way that balances the two, or in REGULATORY PROGRAM a way that minimizes one’s own welfare by sac- rificing one’s life for the other party. In this Consistent with the views of many other re- view, humans have a system that, in each indi- searchers, we have hypothesized that anger is vidual, computes the welfare tradeoff ratio ex- an evolved emotion program with a special re- pressed in the actions of that person toward an- lationship to aggression. However, we think other (individual i to j), and stores it as a that it has an equal relationship to cooperation. summary characterization of i‘s disposition to- In the evolutionary-psychological approach to ward j in the form of a regulatory variable—the the emotions, anger (in addition to being an ex- welfare tradeoff ratio of i to j (WTRij). Indeed, perienced psychological state) is the expression there are at least two parallel, independent wel- of a functionally structured neurocomputa- fare tradeoff ratios: the intrinsic one (intWTR), tional system whose design features and which guides an individual’s behavior toward subcomponents evolved to regulate thinking, another, regardless of whether his or her ac- motivation, and behavior in the context of re- tions are being observed; and the public one solving conflicts of in favor of the an- (publicWTR), which guides an individual’s gry individual (Sell, 2005; Sell, Tooby, & behavior when the recipient (or others) can ob- Cosmides, in press-a, in press-b). Two negotiat- serve the behavior. Some altruism is motivated ing tools regulated by this system are the threat through love, and some through fear, shame, or of inflicting costs (aggression) and the threat of of reward—and the mechanisms involved withdrawing benefits (the down-regulation of are different. cooperation). Humans differ from most other If the human mind really contains welfare species in the number, intensity, and duration tradeoff ratios as regulatory variables that con- of close cooperative relationships, so tradi- trol how well one individual treats another, tional models of animal conflict must be modi- then evolution can build emotions whose func- fied to integrate the cooperative dimension tion is to alter welfare tradeoff ratios in others more fully. Given its apparent functional logic, toward oneself. Anger is conceptualized as a its universality across individuals and cultures mechanism whose functional product is the 132 I. INTERDISCIPLINARY FOUNDATIONS recalibration in the mind of another of this gaining a larger share of disputed resources or other person’s welfare tradeoff ratio with re- social rank. Because strength was consistently spect to oneself. That is, the goal of the system one factor (out of several) relevant under an- (rather than a conscious intention) is to change cestral conditions, and the had the targeted person’s disposition to make wel- reliable access to the body, it seems plausible fare tradeoffs so that he or she more strongly and worth investigating that the mind is de- favors the angered individual in the present and signed to compute a strength self-assessment the future. As in animal contests, the target of automatically and nonconsciously, and to use anger may relinquish a contested resource, or this self-assessment as an input regulating may simply in the future be more careful to behavior. Thus the human brain should have help or to avoid harming the angered individ- evolved a set of programs that (1) evaluates ual. In cooperative relationships, where there is one’s own and other’s formidabilities; (2) trans- the expectation that the cooperative partner forms each of these evaluations into a magni- will spontaneously take the welfare of the indi- tude (a “formidability index”) associated with vidual into account, the primary threat from each person; and, in situations where coopera- the angered person that potentially induces tion is not presumed, (3) implicitly expects or recalibration in the targeted individual is the accords some level of deference based on rela- signaled possibility of the withdrawal of future tive formidability. help and cooperation if the welfare tradeoff ra- The approach briefly sketched above can be tio is not modified. If the withdrawal of this co- unpacked into a large number of empirical pre- operation would be more costly to the target of dictions derived from this analysis of the design the anger than the burden of placing greater features of the program regulating anger. For weight on the welfare of the angry individual, example, it is predicted that in humans, physi- then the target should increase his or her wel- cal strength should be a partial cause of indi- fare tradeoff ratio toward the angry individual, vidual differences in the likelihood of experi- and so treat her or him better in the future. encing and expressing anger. Other things Reciprocally, the program is designed to being equal, stronger individuals are predicted recalibrate the angry individual’s own welfare to be more likely to experience anger and ex- tradeoff ratio toward the target of the anger for press anger; they should feel more entitled; they two functional reasons. This first is that it cur- should expect others to give greater weight to tails the wasteful investment of cooperative ef- their welfare, and become angrier when they fort in individuals who do not respond with a do not. Although physical strength by no sufficient level of cooperation in return. The means exhausts the set of relevant variables, it second is that the potential for this downward offers an easily operationalizable and measur- recalibration functions as leverage to increase able gateway into a series of tests of this gen- the welfare tradeoff ratio of the target toward eral model of the logic underlying the regula- the angry individual. In the absence of coopera- tion of anger. Arguments precipitated by anger tion, the primary threat is the infliction of dam- should reflect the underlying logic of the wel- age. In the presence of cooperation, the pri- fare tradeoff ratio: The complainant will em- mary threat is the withdrawal of cooperation. phasize the cost of the other’s transgression to Concepts that are anchored in the internal reg- him or her, as well as the value of the complain- ulatory variable publicWTR include respect, con- ant’s cooperation to the transgressor, and will sideration, deference, status, rank, and so on. feel more aggrieved if the benefit the transgres- For example, ancestrally, one major cue that sor received (the justification) is small com- an individual would have been able to inflict pared to the cost inflicted. A series of empirical costs to enforce welfare tradeoff ratios in his or studies supports both sets of predictions of this her favor was the individual’s physical strength theory about the design of anger (Sell, 2005; (as noted earlier, we call the ability to inflict Sell et al., in press-a, in press-b). costs “formidability”). Consistent with this, in many species the degree to which an organism values a nonrelative is determined primarily by RECALIBRATIONAL EMOTIONS the relative strength of the two; thus animals SUCH AS GUILT AND GRATITUDE with higher relative strength will, when other factors are held constant, fight more effectively The EEA was full of event relationships (e.g., for resources and have a higher expectation of “Mother is dead”) and psychophysical regular- 8. The Evolutionary Psychology of the Emotions 133 ities (e.g., “Blood indicates injury”) that cued run behavioral response (as fear or anger do), reliable information about the functional but instead to carry out valuation recom- meanings and properties of things, events, per- putations in the light of the new information sons, and regulatory variables to the psycho- relevant to evolved regulatory variables that is logical architecture. For example, certain body provided by external or internal environments proportions and motions indicated immaturity (Tooby & Cosmides, 1990a). An evolutionary and need, activating the emotion program of viewpoint is a utilitarian one, which suggests experiencing cuteness (see Eibl-Ebesfeldt, that the time humans spend simply feeling— 1970). Others indicated sexual attractiveness attending inwardly not to factual representa- (Symons, 1979; Buss, 1994). To be moved with tions, but to something else—is doing some- gratitude, to be glad to be home, to see some- thing useful that will be reflected eventually in one desperately pleading, to hold one’s new- behavior. The hypothesis is that feeling is a born baby in one’s arms for the first time, to see form of computational activity that takes time a family member leave on a long trip, to en- and attention, that can compete with or pre- counter someone desperate with hunger, to empt motivation to engage in other activities, hear one’s baby cry with distress, to be warm and whose function is to recalculate and while it is storming outside—these all mean reweight the regulatory variables implicated by something to us. How does this happen? In ad- the newly encountered information. This ap- dition to the situation-detecting algorithms as- proach has the potential to provide an account sociated with major emotion programs such as of the characteristics of emotions such as guilt fear, anger, or jealousy, we believe that humans or depression, which appear otherwise puz- have a far larger set of evolved specializations, zling from a functional perspective. The feel- which we call “recalibrational releasing en- ings these emotion programs engender interfere gines.” These are activated by situation- with short-term utilitarian action that an active detecting algorithms, and their function is to organism might be expected to engage in. If trigger appropriate recalibrations, including they were not useful, the capacity to feel them affective recalibrations, when certain would have been selected out. evolutionarily recognizable situations are en- Consider guilt: We believe that guilt func- countered. tions as an emotion mode specialized for We believe that the psychophysical or inter- recalibration of regulatory variables that con- pretive “front ends” of emotion programs use trol tradeoffs in welfare between self and other these cues not only to trigger the appropriate (Tooby & Cosmides, 1990a). Three important emotion, but to alter the weightings of regula- reasons why humans evolved to take the wel- tory variables embedded in decision rules. (For fare of others into account are genetic related- example, if you experience someone ness toward relatives (Hamilton, 1964), the you disrespectfully, it makes you angry.) In- positive externalities others emit (Tooby & deed, most evolutionarily recurrent situations Cosmides, 1996a), and the maintenance of co- that selected for corresponding emotion pro- operative relationships (Trivers, 1971; Tooby grams bristle with information that allows the & Cosmides, 1996a). The regulatory variable recomputation of one or more variables. approach provides a clear framework for un- Recalibration (which, when consciously acces- derstanding why guilt evolved and what its un- sible, appears to produce rich and distinct feel- derlying logic is. In this view, guilt involves the ing states) is therefore a major functional com- recalibration of regulatory variables considered ponent of most emotion programs. Jealousy, when one is making decisions about tradeoffs for example, involves several sets of in welfare between the self and others, based recalibrations (e.g., diminution in estimate of on new information about actual or potential one’s own mate value, diminution of , low- harm arising from having placed too little ering of the welfare tradeoff ratio toward the weight on the other person’s welfare in past ac- mate). tions. would favor a mechanism Indeed, from an evolutionary-psychological designed to effect such recalibration toward perspective, recalibrational emotion programs those the kin detection mechanism identifies as appear to be the dominant (but not the only) close genetic relatives. Similarly, individuals components of such emotions as guilt, grief, have an intrinsic interest in the welfare of those depression, shame, and gratitude. Their pri- whose existence benefits them, and with whom mary function is not to orchestrate any short- they share deep relationships 134 I. INTERDISCIPLINARY FOUNDATIONS

(Tooby & Cosmides, 1996a). Third, reciprocal, When guilt is triggered, the welfare tradeoff ra- exchange, or cooperative relationships need to tio is adjusted, as well as a variety of subsidiary be proximately motivated, so that benefit flows variables expressing this ratio in action. As a are appropriately titrated. Individuals who ex- result of this recalibration, the guilty individ- perienced guilt (and the associated modifica- ual’s behavior should reflect this higher valua- tion of decision rules) would have been less tion. In cases where the effects were inten- likely to injure relationship partners repeatedly, tional and anticipated, there should be little and they would have had more success in main- recalibration. taining beneficial cooperative relationships. Existing findings substantiate these predic- In the case of kin selection, we now have an tions and explain some of their otherwise puz- empirical map of the architecture of the zling features. Unsurprisingly, when the valued neurocomputational program that detects ge- other is negatively affected unexpectedly, sub- netic relatedness and passes this information to jects feel guiltier (Baumeister, Stillwell, & the welfare tradeoff system (Lieberman et al., Heatherton, 1995, Kubany & Watson, 2003). 2007). The theory of kin selection says noth- More surprisingly, individuals feel guiltier ing, however, about the procedures by which a when the harm was caused accidentally rather mechanism could estimate the value of, say, a than anticipated, even though individuals are particular piece of food to oneself and one’s usually considered less responsible and culpa- kin. The fitness payoffs of such acts of assis- ble for the harm when it occurs accidentally tance vary with circumstances. Consequently, (McGraw, 1987; Baumeister et al., 1995). If the each decision about where to allocate assis- function of guilt is, however, to recalibrate an tance depends on inferences about the relative improperly set welfare tradeoff ratio, then in- weights of these variables. These nonconscious formation that merely confirms the evaluation computations (however they are carried out) present in the decision requires no recal- must be subject to error, selecting for feedback ibration. If the effect was foreseen and chosen systems of correction. anyway in the light of the existing ratio, then Imagine a hunter–gatherer woman with a no adjustment is necessary. sister. The mechanisms in the woman’s brain Gratitude is a recalibrational emotion pro- have been using the best information available gram that is complementary to guilt. Guilt to her to weight the relative values of the turns up the welfare tradeoff ratio toward an she has been acquiring to herself and her sister, individual when one has evidence that one’s leaving her reassured that it is safe to leave her own actions have expressed too low a valua- sister for a while without provisioning her. The tion of the other. Gratitude is triggered by new sudden discovery that her sister, since she was information indicating that another places a last contacted, has been starving and has higher value on one’s welfare than one’s system become desperately sick functions as an had previously estimated—again leading to an information-dense situation allowing the up-regulation of the WTR toward that person. recalibration of the algorithms weighting the Anger, guilt, and gratitude all play different relative values of the meat to self and sister roles in cooperation, and their computational (among other things). The sister’s sickness structure reflects their recalibrational functions functions as a cue that the previous allocation with respect to welfare tradeoff ratios and the weighting was in error and that the variables choice points they involve. need to be reweighted—including all of the weightings embedded in habitual action se- The evolutionary-psychological stance moti- quences that might be relevant to the sister’s vating the investigation of the program archi- welfare. Guilt should be triggered when the in- tecture of the emotions suggests that the emo- dividual receives (1) unanticipated information tions are intricate, functionally organized, and about the welfare of a valued other (or the in- sensitively related to the detailed structure of creased value of the other), indicating that (2) ancestral problems. In this view, the emotions the actor’s actions or omissions caused or al- are likely to be far more sophisticated engineer- lowed the welfare of the valued individual to be ing achievements than previously appreciated, damaged in a way that is inconsistent with the and there are many decades of work ahead for actor’s welfare tradeoff ratio, given (3) the emotion researchers before they are compre- actor’s resources and potential for action. hensively mapped. 8. The Evolutionary Psychology of the Emotions 135

NOTES Brown, D. (1991). Human universals. New York: McGraw-Hill. 1. Marks (1987, pp. 68–69) vividly conveys how many Boyer, P., & Liénard, P. (2006). Why ritualized behav- aspects of behavior and physiology may be entrained ior? Precaution systems and action parsing in devel- by certain kinds of fear: opmental, pathological, and cultural rituals. Behav- ioral and Brain Sciences, 12(6), 595–613. During extreme fear humans may be “scared stiff” or Buss, D. M. (1994). . New York: “frozen with fear.” A paralyzed conscious state with . abrupt onset and termination is reported by survivors Buss, D. M. (Ed.). (2005). The handbook of evolution- of attacks by wild animals, by shell-shocked soldiers, and by more than 50% of rape victims (Suarez & Gal- ary psychology. Hoboken, NJ: Wiley. lup, 1979). Similarities between tonic immobility and Cannon, W. (1929). Bodily changes in pain, hunger, fear rape-induced paralysis were listed by Suarez & Gallup and rage: Researches into the function of emotional (features noted by rape victims are in parentheses): (1) excitement. New York: Harper & Row. profound motor inhibition (inability to move); (2) Chagnon, N. (1983). Yanomamo: The fierce people (3rd Parkinsonian-like tremors (body-shaking); (3) silence ed.). New York: Holt, Rinehart & Winston. (inability to call out or scream); (4) no loss of con- Cosmides, L. (1989). The logic of social exchange: Has sciousness testified by retention of conditioned reac- natural selection shaped how humans reason? tions acquired during the immobility (recall of details of the attack); (5) apparent analgesia (numbness and Studies with the . Cognition, 31, insensitivity to pain); (6) reduced core temperature 187–276. (sensation of feeling cold); (7) abrupt onset and termi- Cosmides, L., & Tooby, J. (1987). From evolution to nation (sudden onset and remission of paralysis); (8) behavior: Evolutionary psychology as the missing aggressive reactions at termination (attack of the rap- link. In J. Dupre (Ed.), The latest on the best: Essays ist after recovery); (9) frequent inhibition of attack by on evolution and optimality (pp. 276–306). Cam- a predator . . . bridge, MA: MIT Press. Cosmides, L., & Tooby, J. (1989). Evolutionary psy- 2. We think that some emotion programs evolved in re- chology and the generation of culture: Part II. Case sponse to the situation cue provided by a strong drive study: A computational theory of social exchange. state, such as hunger, when the motivational intensity and , 10, 51–97. reached a point that other mechanisms became domi- Cosmides, L., & Tooby, J. (1992). Cognitive adapta- nated and entrained by the magnitude of the motiva- tions for social exchange. In J. Barkow, L. Cosmides, tion. We see no principled reason for distinguishing & J. Tooby (Eds.), : Evolutionary strong drive states from other emotion programs, psychology and the generation of culture (pp. 163– and suspect that this practice originated from out- 228). New York: . dated notions of natural selection that separated Cosmides, L., & Tooby, J. (1997). Dissecting the com- “survival-related” functions (hunger, thirst) from putational architecture of social inference mecha- other functions, such as mate acquisition or reciproc- nisms. In G. Bock & G. Cardeco (Eds.), Character- ity. Thus we propose that it is useful to model special- izing human psychological adaptations (Ciba ized motivational states as emotion programs, just as Symposium No. 208, pp. 132–156). Chichester, UK: one would disgust, anger, or fear. Wiley. 3. The evolutionary purpose of deceitful emotional ex- Cosmides, L., & Tooby, J. (2000). Evolutionary psy- pressions is to (falsely) communicate the same two chology and the emotions. In M. Lewis & J. M. things. Haviland-Jones (Eds.), Handbook of emotions (2nd ed., pp. 91–115). New York: Guilford Press. REFERENCES Crawford, C., & Krebs, D. (Eds.). (1998). Handbook of evolutionary psychology. Mahwah, NJ: Erlbaum. Archer, J. (1988). The behavioural biology of aggres- Daly, M., & Wilson, M. (1988). Homicide. New York: sion. Cambridge, UK: Cambridge University Press. Aldine. Barkow, J., Cosmides, L., & Tooby, J. (Eds.). (1992). Daly, M., Wilson, M., & Weghorst, S. J. (1982). Male The adapted mind: Evolutionary psychology and the sexual jealousy. Ethology and Sociobiology, 3, 11– generation of culture. New York: Oxford University 27. Press. Derryberry, D., & Tucker, D. (1994). Motivating the fo- Baron-Cohen, S. (1995). Mindblindness: An essay on cus of attention. In P. M. Neidenthal & S. Kitayama autism and . Cambridge, MA: MIT (Eds.), The heart’s eye: Emotional influences in per- Press. ception and attention (pp. 167–196). San Diego, CA: Baumeister, R., Stillwell, A., & Heatherton, T. (1995). Academic Press. Personal narratives about guilt. Basic and Applied Eibl-Ebesfeldt, I. (1970). Ethology: The biology of , 17(1–2), 173–198. behavior. New York: Holt, Rinehart & Winston. Berkow, R. (Ed.). (1992). The Merck manual of diagno- Ekman, P. (1973). Cross-cultural studies of facial ex- sis and therapy (16th ed.). Rahway, NJ: Merck. pression. In P. Ekman (Ed.), Darwin and facial ex- 136 I. INTERDISCIPLINARY FOUNDATIONS

pression: A century of research in review (pp. 169– attribution of responsibility approach. Journal of 222). New York: Academic Press. Personality and Social Psychology, 53(2), 247–256. Ekman, P. (Ed.). (1982). Emotion in the human face Miller, S. (Ed.). (1976). Symptoms: The complete home (2nd ed.). Cambridge, UK: Cambridge University medical encyclopedia. New York: Crowell. Press. Mineka, S., & Cook, M. (1993). Mechanisms involved Ekman, P. (1985). Telling lies. New York: Norton. in the observational conditioning of fear. Journal of Ekman, P. (1994). Strong evidence for universals in fa- : General, 122, 23–38. cial expressions. Psychological Bulletin, 115, 268– Mowat, R. R. (1966). Morbid jealousy and : A 287. psychiatric study of morbidly jealous at Ekman, P., Levenson, R., & Friesen, W. (1983). Auto- Broadmoor. London: Tavistock. nomic nervous system activities distinguishes be- Nesse, R. (1991). Evolutionary explanations of emo- tween emotions. Science, 221, 1208–1210. tions. , 1, 261–289. Fiddick, L., Cosmides, L., & Tooby, J. (2000). No inter- Pinker, S. (1997). . New York: pretation without representation: The role of Norton. domain-specific representations and inferences in the Pitman, R., & Orr, S. (1995). of emo- Wason selection task. Cognition, 77, 1–79. tional and memory networks in posttraumatic stress Frank, R. (1988). Passions within reason: The strategic disorder. In J. McGaugh, N. Weinberger, & G. Lynch role of the emotions. New York: Norton. (Eds.), Brain and memory: Modulation and media- Fredrickson, B. (1998). What good are positive emo- tion of neuroplasticity (pp. 75–83). New York: Ox- tions? Review of General Psychology, 2, 300–319. ford University Press. Fridlund, A. (1994). Human facial expression: An evo- Raymond, E. S. (1991). The new hacker’s dictionary. lutionary view. San Diego, CA: Academic Press. Cambridge, MA: MIT Press. Garcia, J. (1990). Learning without memory. Journal of Sell, A. (2005). Regulating welfare trade-off ratios: , 2, 287–305. Three tests of an evolutionary–computational model Gigerenzer, G., & Murray, D. (1987). Cognition as intu- of human anger. Doctoral dissertation, Department itive statistics. Hillsdale, NJ: Erlbaum. of Psychology, University of California, Santa Gould, S. J. (1997, June 12). Darwinian fundamental- Barbara. ism. New York Review of Books, pp. 34–38. Sell, A., Tooby, J., & Cosmides, L. (in press-a). The logic Hamilton, W. D. (1964). The genetical evolution of so- of anger: Formidability regulates human conflict. cial behaviour. Journal of Theoretical Biology, 7,1– Sell, A., Tooby, J., & Cosmides, L. (in press-b). Compu- 52. tation and the logic of anger. Haselton, M. G., & Buss, D. M. (2003). Biases in social Shepard, R. N. (1987). Evolution of a mesh between judgment: Design flaws or design features? In J. principles of the mind and regularities of the world. Forgas, K. Williams, & B. von Hippel (Eds.), Re- In J. Dupre (Ed.), The latest on the best: Essays on sponding to the social world: Implicit and explicit evolution and optimality (pp. 251–275). Cambridge, processes in social judgments and decisions (pp. 23– MA: MIT Press. 43). New York: Cambridge University Press. Shepherd, M. (1961). Morbid jealousy: Some clinical Hirshleifer, J. (1987). On the emotions as guarantors of and social aspects of a psychiatric symptom. Journal threats and promises. In J. Dupre (Ed.), The latest on of Mental Science, 107, 687–753. the best: Essays on evolution and optimality (pp. Stenberg, C. R., & Campos, J. J. (1990). The develop- 307–326). Cambridge, MA: MIT Press. ment of anger expression in infancy. In N. Stein, B. Kubany, E., & Watson, S. (2003). Guilt: Elaboration of Leventhal, & T. Trabasso (Eds.), Psychological and a multidimensional model. Psychological Record, 53, biological approaches to emotion (pp. 247–282). 51–90. Hillsdale, NJ: Erlbaum. Lazarus, R., & Lazarus, B. (1994). and reason. Stenberg, C. R, Campos, J. J., & Emde, R. N. (1983). New York: Oxford University Press. The facial expression of anger in seven-month-old in- LeDoux, J. (1995). In search of an emotional system in fants. , 54(1), 178–184. the brain: Leaping from fear to emotion to conscious- Suarez, S. D., & Gallup, G. G. (1979). Tonic immobility ness. In M. S. Gazzaniga (Ed.), The cognitive neuro- as a response to rage in humans: A theoretical note. sciences (pp. 1049–1061). Cambridge, MA: MIT Psychological Record, 29, 315–320. Press. Swets, J. A., Tanner, W. D., & Birdsall, T. G. (1961). De- Lieberman, D., Tooby, J., & Cosmides, L. (2007). The cision processes in perception. , architecture of human kin detection. Nature, 445, 68, 301–340. 727–731. Symons, D. (1979). The evolution of . Marks, I. (1987). , phobias, and rituals. New York: Oxford: Oxford University Press. Oxford University Press. Tomaka, J., Blascovich, J., Kibler, J., & Ernst, J. (1997). Maynard Smith, J. (1982). Evolution and the theory of Cognitive and physiological antecedents of threat games. Cambridge, UK: Cambridge University Press. and challenge appraisal. Journal of Personality and McGraw, K. (1987). Guilt following transgression: an Social Psychology, 73, 63–72. 8. The Evolutionary Psychology of the Emotions 137

Tooby, J. (1985). The emergence of evolutionary psy- Tooby, J., & Cosmides, L. (1997). On evolutionary psy- chology. In D. Pines (Ed.), Emerging syntheses in sci- chology and modern : A reply to Ste- ence: Proceedings of the founding workshops of the phen Jay Gould. Available online at www.psy- Santa Fe Institute (pp. 1–6). Santa Fe, NM: Santa Fe ch.ucsb.edu/research/cep/critical_eye.htm Institute. Tooby, J., & Cosmides, L. (2005). Conceptual founda- Tooby, J., & Cosmides, L. (1990a). The past explains tions of evolutionary psychology. In D. M. Buss the present: emotional adaptations and the structure (Ed.), The handbook of evolutionary psychology (pp. of ancestral environments. Ethology and Socio- 5–67). Hoboken, NJ: Wiley. biology, 11, 375–424. Tooby, J., & Cosmides, L. (in press). Internal regulatory Tooby, J., & Cosmides, L. (1990b). On the universality variables and the design of human motivation: An of human nature and the uniqueness of the individ- evolutionary and computational approach. In A. ual: The role of genetics and adaptation. Journal of Elliot (Ed.), Handbook of approach and avoidance Personality, 58, 17–67. motivation. Mahwah, NJ: Erlbaum. Tooby, J., & Cosmides, L. (1992). The psychological Tooby, J., Cosmides, L., & Barrett, H. C. (2003). The foundations of culture. In J. Barkow, L. Cosmides, & second of is the first law of psy- J. Tooby (Eds.), The adapted mind: Evolutionary chology: Evolutionary psychology and the generation of culture (pp. 19– and the theory of tandem, coordinated inheritances. 136). New York: Oxford University Press. Psychological Bulletin, 129(6), 858–865. Tooby, J., & Cosmides, L. (1996a). Friendship and the Tooby, J., Cosmides, L., Sell, A., Lieberman, D., & banker’s paradox: Other pathways to the evolution Sznycer, D. (in press). Internal regulatory variables of adaptations for altruism. In W. G. Runciman, J. and the design of human motivation: An evolution- M. Smith, & R. I. M. Dunbar (Eds.), Evolution of ary and computational approach. In A. Elliot (Ed.), Social Behaviour Patterns in and Man. Pro- Handbook of approach and avoidance motivation. ceedings of the British Academy, 88, 119–143. Mahwah, NJ: Erlbaum. Tooby, J., & Cosmides, L. (1996b). The computa- Trivers, R. L. (1971). The evolution of reciprocal altru- tionalist theory of communication. Paper presented ism. The Quarterly Review of Biology, 46, 35–37. at the meeting of the and Evolu- Williams, G. C. (1966). Adaptation and natural selec- tion Meetings, University of Arizona, Tuc- tion: A critique of some current evolutionary son. thought. Princeton, NJ: Princeton University Press.