For the good that I would I do not: but the evil which I would not, that I do.
—Romans 7.19, King James Version
In the chapter on will from his Principles of Psychology, William James discussed five types of decisions. Most decisions he noted were decisions without effort, but in the
final type of decision, the feeling that the evidence is all in, and that reason has balanced the books, may be either present or absent. But in either case we feel, in deciding, as if we ourselves by our own willful act inclined the beam: . . . If examined closely, its chief difference from the former cases appears to be that in those cases the mind at the moment of deciding on the triumphant alternative dropped the other one wholly or nearly out of sight, whereas here both alternatives are steadily held in view, and in the very act of murdering the vanquished possibility the chooser realizes how much in that instant he is making himself lose. It is deliberately driving a thorn into one’s flesh; and the sense of inward effort with which the act is accompanied is an element which sets the fifth type of decision in strong contrast with the previous four varieties, and makes of it an altogether peculiar sort of mental phenomenon. (1890/1983, 1141)
After consideration of the kinds of decisions that were made with and without effort, James concluded that “effort complicates volition . . . whenever a rarer and more ideal impulse is called upon to neutralize others of a more instinctive and habitual kind.”
Religious, literary, and psychoanalytic texts abound with discussions of conflicts between our higher and lower natures, between passion and reason, between selfishness and selflessness, between immediate gratification and pursuit of long-term goals. We all are familiar with being caught on the horns of a dilemma, of wanting to make a phone call and simultaneously not wanting to make the call, of being torn between temptation and conscience; but evolutionary biology has had little to say on why our subjective experience should be organized in this manner. At first sight, the idea that we can be at war with ourselves appears paradoxical. If we are products of natural selection, superbly designed to maximize inclusive fitness, why do we often find it hard to make decisions and stick to them? A fitness-maximizing computer would simply calculate the expected utilities of the different alternatives and then choose the alternative with the highest motivational score. Why should some kinds of decisions be more difficult to make than others? Is the subjective experience of effort merely a measure of the computational complexity of a problem, or is something else going on?
For William James, “The existence of the effort as a phenomenal fact in our consciousness cannot of course be doubted or denied. Its significance, on the other hand, is a matter about which the gravest difference of opinion prevails. Questions as momentous as that of the very existence of spiritual causality, as vast as that of universal predestination or free-will, depend on its interpretation” (1890/1983, 535). My aim in this chapter is not to address such momentous questions, nor to shed light on really difficult questions such as how and why we have subjective experiences. Rather, it is to ask how one might begin to reconcile nonbiologists’ perception of the ubiquity of internal conflict with biologists’ view of the mind as an adaptive product of natural selection. Internal conflict often seems maladaptive; consuming time, energy, and repose. If so, why does it persist?
Three kinds of hypotheses could resolve the conundrum of conflict within an adapted mind. First, one might argue that internal conflict arises from constraints on the perfection of adaptation; that evolved mechanisms work well on average but occasionally malfunction. We would be better off without internal conflict, but we are stuck with it. Second, one might argue that internal conflict is adaptive and the “contending parties” have the same ultimate ends and their conflict is in a sense “illusory.” Natural selection has simply adopted an adversarial system as the best mechanism of arriving at useful truths. Finally, one might argue that internal conflict is “real” and reflects a disagreement over ultimate ends among multiple agents that contribute to mental activity. I will reveal my hand at the outset. I believe that all three kinds of explanations, together with their complex interactions, contribute to experiences of internal conflict.
First, let me briefly consider nonadaptive interpretations of intrapersonal conflicts. The precision of achievable adaptation is limited because natural selection is retrospective, fitting us to the past rather than the present environment; because the adaptive response to environmental change is limited by the pool of available variation and by time-lags until the origin of appropriate new mutations; and because selection is blind to very weak selective forces (i.e., chance plays a large role in who survives and reproduces when differences in adaptedness are slight). Some internal conflicts may merely reflect the imprecision of adaptation. Our genomes evolve by minor revisions to an old text just as the operating systems of computers evolve by the addition of new functionalities to old code; neither programmers nor natural selection have been able to eliminate all opportunities for malfunction. On this view, some internal conflicts may be analogous to “system conflicts” that occasionally cause my computer to crash: multiple functional programs are running simultaneously and occasionally make contradictory or ambiguous demands on the operating system. The analogy is, of course, limited. My computer does not, in fact, run multiple programs simultaneously. Instead, it is a serial machine that has only a single program running in its central processor at any particular moment, but switches rapidly between programs. Our brains, by contrast, are massively parallel processors with different subsystems handling different kinds of data. Somehow this dispersed neural activity has to be integrated in coming to a decision. Perhaps “conflict” could arise from imperfections in the process of integration.
Without doubt, our current environment presents us with novel challenges for which we lack specific adaptations. There were no opportunities in our evolutionary past to put aside resources for ten or twenty years, and then recover them with interest. Retirement planning is a recent cultural innovation for which we are unlikely to have evolved dedicated mechanisms. Instead, we employ general-purpose problem-solving machinery to make plans that come into conflict with more hardwired responses. My rational resolution to save is thwarted by short-term impulses that fritter away income on ephemeral goods. (In this case, it is unclear that a comfortable retirement has anything to do with enhancing fitness. From a genetic perspective, our impulses may have it right.) Powerful narcotics are another novelty for which we are adaptively unprepared. An addict may strongly desire to be free of his compulsion, but may lack the will to override strongly maladaptive, albeit evolutionarily programmed, cravings.
Adaptive explanations of internal conflict often invoke competition among alternatives as the most effective mechanism of choosing the best course of action. Consider a gazelle who is racing toward a stump before which he must either zig to the right or zag to the left. As the decision point approaches, the gazelle calculates probable outcomes if he zigs or he zags, in response to new information about the terrain to either side and the fear from which he flees. These plans of action compete against each other until he abruptly commits to one of the alternatives. The cheetah, in hot pursuit, also needs to plan ahead for both eventualities, to respond rapidly once the gazelle commits but not be fooled by a feint to one side. There is a premium on minimizing reactions times. The gazelle commits to one of his options and the cheetah responds with one of hers. The gazelle’s options for zigging or zagging are in “conflict” until action, but “at the moment of deciding” the rejected alternative is dropped wholly from view. Such a model seems inadequate to explain “driving a thorn into one’s flesh.”
In an earlier chapter on instinct, James had considered conflicts of impulses that arose in one of two ways. First, an individual might have instinctive impulses to respond to O with A and P with B, but O could become a sign of P from experience, “so that when he meets O the immediate impulse A and the remote impulse B struggle in his breast for the mastery” (1011). Second, “Nature implants contrary impulses to act on many classes of things, and leaves it to slight alterations in the conditions of the individual case to decide which impulse shall carry the day” (1013). In both these models, James envisioned experience as deciding between contradictory impulses. The adjudication could be partly mediated by reason: “Reason, per se, can inhibit no impulses; the only thing that can neutralize an impulse is an impulse the other way. Reason may, however, make an inference which will excite the imagination so as to let loose the impulse the other way” (1013). This model of conflicting impulses resolved by experience can be likened to a court with a disinterested judge but, although James writes of a struggle for mastery between impulses, it is not immediately clear why such a process should be accompanied by feelings of effort.
Humans have evolved general-purpose problem-solving mechanisms (reason), and the ability to learn from others, to compensate for the limitations of hardwired instincts. We are rational, cultural, and instinctive beings. Sometimes these alternative sources of behavioral guidance promote different choices. Instinct summarizes the wisdom of past natural selection and recommends actions that have worked before under similar circumstances. Culture also summarizes wisdom from the past and can respond much faster than gene sequence to environmental change, but, from a gene’s-eye view, has the disadvantage of evolving by rules that need not promote genetic fitness. Reason can respond to unique features of the current situation, and to weak selective forces, but may lack the historical judgment of either instinct or culture. Our passions, both positive and negative, are the carrots and sticks employed by genes to mold our actions to their ends. Reason may be a slave to the passions, but reason pursues pleasures as ends in themselves rather than as means to an end. (Coitus with rubber is an obvious example of reason circumventing genes’ ends.)
Our ability to reason is an adaptive response to the imperfections of instinct, but this adaptation must have its own imperfections, including inevitable clashes when instinct and reason offer conflicting advice. But if such clashes are unavoidable and recurrent, humans should have evolved adaptive (although imperfect) mechanisms of resolving them. How might a well-designed organism resolve conflicts between the dictates of instinct and reason? Such an organism might have a limited ability to override instinct given strong enough reasons, with “strong enough” calibrated to match the strength of past selection favoring the instinctive response. Very strong motivation would be needed for reason to prevail in decisions closely related to fitness, for which instinct provides a powerful guide, but the threshold of motivation could be lowered when the prescripts of instinct are less strong. These considerations suggest an adaptive explanation for the feeling of effort in making certain kinds of decisions. Some decisions are hard to make because individuals in the past who made similar decisions with greater ease left fewer offspring. The strength of our will, no less than the strength of our muscles, can be shaped by natural selection. Moralists may derive some comfort from the muscular response to exercise.
Reality is more complicated than the above naïve model. Instincts are not unitary; neither is reason (itself a special kind of instinct). Different parts of the brain undertake different tasks, and no part has access to the big picture. An arrangement in which mental modules compete for attention and influence could be a general organizing principle of the mind. Different modules process different kinds of data to produce recommendations for action. What a module communicates to the decision-making collective may be no more than its preference rather than a detailed justification. The preferences must then be aggregated to generate a choice. Kenneth Arrow (1963) proved that no procedure for aggregating preferences in situations of social choice can be guaranteed never to violate basic axioms of rationality. His proof assumed that information was limited to the rank-ordering of preferences, and specifically did not allow interpersonal comparisons of the strength of preferences. Are there similar constraints on the rationality of intrapersonal aggregation of preferences?
In my subjective experience, choice becomes more difficult when motivations are not expressed in a common currency. Life would be simpler if one could simply compare expected degrees of pleasure from giving way to a sexual infidelity and from resisting it. Instead, the two courses of action promise rewards that differ in kind. There may be functional reasons for having more than one currency of reward—some rewards could be more suited to sustaining long-term projects, others for providing immediate gratification—but their existence renders intrapersonal comparisons problematic. Multiple currencies would still allow facile comparisons among rewards if their exchange rates were well defined, but this does not seem to be the case. Why should this be so? If one part of my mind counsels one action and another part counsels another, how is this disagreement to be resolved if there is no common unit for comparison?
Perhaps there are benefits, as well as costs, to not expressing all values on a single scale. Multiple currencies allow exchange rates to vary. This might be adaptive if the optimal weights to be attached to the pleadings of different internal voices vary from place to place and time to time. A woman who yields to an extramarital passion may face very different consequences in New York and Riyadh. Multiple currencies could allow her to learn the appropriate exchange rates for her culture from her experience of which choices were rewarded and which punished during her behavioral maturation. If so, choices might be expected to become easier as we grow older and learn the norms of our culture (assuming these norms are not rapidly changing).
The discussion so far has concerned questions of adaptation and constraint: there may be adaptive reasons for allowing internal voices to compete for attention, but some expressions of this competition may be maladaptive. No mechanism is perfect. Pathological indecision may be just that: pathological. A model of the mind has been presented in which mental modules may express different preferences because they have different capabilities and process different inputs. In this model, the problem of preference aggregation reduces to a problem of determining the best weights to assign to the preferences of different modules. However, another possibility should be considered: internal factions may disagree about ultimate ends. Each faction may have an incentive to overstate the case for its side of the argument, even for broadcasting misleading information.
Agents with different interests may have different preferences despite access to the same information. If the self is an assemblage of agents with distinct interests, then internal conflict may reflect disagreements over ultimate ends. What benefits one, need not benefit all. Factions may disagree about the weights that should be assigned to the recommendations of different parliamentary committees. There need be no agreement on exchange rates, even when cultural norms have been fully internalized. In this view, decision making would resemble the deliberations of a collective: sometimes consensus is achieved, sometimes one set of interests overrules the others, and sometimes the committee fails to decide.
I have in mind two kinds of agents with stakes in the deliberations of self. The first are genes. Genes’ ultimate ends are the propagation of their copies. Put another way, the genes that we see today are those that have successfully propagated in the past. Genes can be said to have purposes to the extent that they possess properties that have promoted their own survival and replication. The second are ideas (or, to use Richard Dawkins’s term, memes). Ideas may be propagated from other minds or generated afresh within a mind. (In fact, most of our ideas are hybrids that recombine content acquired from other minds with features generated in our own mind.) The ideas that reach our awareness have succeeded in competition with other ideas for attention. The ideas that colonize other minds have succeeded in competition with other ideas for expression by the transmitter and in competition with other ideas for perception by the receiver. Ideas can be said to have purposes to the extent that they possess properties that have promoted their propagation from mind to mind. Such properties are the “adaptations” of ideas.
Ideas compete for space on the page. There has been conflict in my mind over the writing of this chapter. I have often been undecided about what to write and this indecision has preoccupied my mind, crowding out other concerns. Different ideas and forms of expression have competed for expression. Many sentences were written, only to be erased and replaced by another sentence that I had temporarily discarded. Slowly a final text took form. What were the properties that made an idea successful in this competition? One of them was coherence with the rest of the manuscript. Another was my estimate of how likely an idea was to attract your attention. I preferred pithy to dull formulations. Another criterion, I hope, was some degree of correspondence with reality, what one might call “truth.” Now some of these ideas have entered your mind, gentle reader. May they go forth and multiply.
Why do I care whether my book is read? Why do people, in general, care about the propagation of their ideas? The fact that we care suggests there has been an evolutionary correlation between the transmission of ideas and of genes. In other words, successful propagators of ideas have, on average, also been successful propagators of genes. Ideas can be useful, helping oneself and one’s kin to survive in a challenging physical environment. And ideas can function in display: I want to impress you with how smart I am; I do not want to say anything silly; I want my ideas to spread so that I can bask in their reflected glory. Being a propagator of successful ideas must, on average, have translated into influence and control of resources. We have an instinct to generate and propagate ideas, and this instinct creates the environment in which ideas can compete and evolve their own purposes.
“Good” ideas are persuasive. They appeal to our genetic biases, even if they do not promote genetic fitness. The correlation between genetic and memetic propagation cannot have been perfect, and, at times, ideas may have propagated at the expense of genes, or genes at the expense of ideas. When a person willingly dies for a political or religious ideal, the incitements of ideas have triumphed over the urgings of genes. But, when a charismatic preacher compromises his ministry for a brief sexual encounter, the entreaties of his genes have trumped those of ideas. Natural selection favors the evolution of genetic biases against adopting ideas that reduce genetic fitness, but ideas evolve much faster than genes and will accommodate themselves to these new biases. Our choices are shaped by the interplay of the stubborn intransigence of genes and the supple agility of ideas.
But surely, one might object, we should evolve a consistent set of genetic biases. Not necessarily, if we are also subject to conflict among genes. Biologists commonly assume that all genes of an individual have the same interests because all have the same chance of being transmitted to that individual’s progeny. For most choices, this is probably a reasonable assumption. However, there are subtle ways in which genes can have distinct interests, and these can promote contradictory adaptations within the genome. Transposable elements replicate faster than the rest of the genome. Nuclear genes are transmitted via eggs and sperm; mitochondrial genes are transmitted only via eggs. If different genes have different rules of transmission, then an adaptation that promotes the long-term propagation of a gene may not promote the transmission of the other genes with which it is temporarily associated. In the remainder of this chapter, I will concentrate on the specific case of conflict between the nuclear genes we inherit from our fathers and those we inherit from our mothers, but I will first digress on how natural selection acts on interactions among relatives.
We all die. A gene in a neuron does not leave direct descendants. Only genes that are present in our sperm or eggs have any chance of producing copies that survive our death. Nevertheless, each gene in the neuron is a copy of a gene in the fertilized egg that gave rise to both our brains and our gonads, and each gene in the egg has left its copies in both organs. For this reason, genes in brains have evolved complex adaptations that promote the propagation of their indirect copies in gonads.
Our body contains multiple kinds of cells, all with the same set of genes but with different roles to play in facilitating the transmission of gene copies via our gonads. If a gene in my liver can promote the transmission of its indirect copies in my gonads, then there is no reason why it cannot also promote the transmission of its indirect copies in the gonads of my relatives. When a mother provides milk to her baby, the genes in her breast promote the propagation of gene copies in the baby’s gonads, not the mother’s gonads. Lactation is associated with temporary infertility of the mother, delaying the birth of her next child. Thus, nursing promotes the propagation of gene copies in the baby’s gonads at the expense of gene copies in the mother’s gonads. There is no guarantee that a gene in the mother’s breast has indirect copies in a particular baby. Rather, a gene in her breast has one chance in two of having an indirect copy transmitted to the baby. (The mother receives half her genes from her mother and half from her father. She transmits half to her baby, but this half is a mixture of genes she received from her mother and father.)
Among cooperatively breeding meerkats (a kind of mongoose), daughters sometimes help their mothers by nursing younger siblings. By this means, genes in the daughter’s breast may promote the propagation of their indirect copies in the mother’s gonads, because the mother is relieved of some of the costs of lactation, and in the sibling’s gonads, because the sibling is the direct beneficiary of extra milk. Each gene in a daughter’s breast has one chance in two of having indirect copies in the gonads of a sibling receiving extra milk, the same as the probability that the gene would have indirect copies in the gonads of one of the daughter’s own offspring. If genes in breasts gain equivalent benefits from providing milk to offspring or siblings, why is sororal suckling not more common among mammals?
The equivalence of these two routes of promoting genetic fitness is predicated on the assumption that donor and recipient are full siblings, sharing the same mother and father, as is usually the case in meerkats. Suppose, however, that the younger sibling has the same mother but a different father. A gene that a daughter inherits from her mother has one chance in two of having an indirect copy in the younger sibling and one chance in two of having an indirect copy in the daughter’s own offspring. The two routes of achieving fitness remain equivalent for genes of maternal origin. However, a gene that the daughter inherits from her father is absent from the younger sibling but has one chance in two of having an indirect copy transmitted to the daughter’s own offspring. The two routes are not equivalent for genes of paternal origin. That is, maternal genes of the daughter are “indifferent” as to whether an offspring or a maternal half-sibling receives a benefit, but the daughter’s paternal genes would “prefer” her own offspring to receive a benefit instead of a maternal half-sibling. Thus, under some circumstances maternal and paternal genes in a daughter’s breast could “disagree” over whether to supply milk to a half-sibling.
The above simple example illustrates the general point that most relatives, excepting our direct descendants and full siblings, are genetic kin of our mother or father but not of both. Genes of paternal origin should “care” about their effects on the father’s side of the family, but not the mother’s side, whereas the situation is reversed for genes of maternal origin. Genes of maternal and paternal origin thus occupy different “social environments” and may evolve different behaviors suited to their different circumstances. A biblical example can illustrate asymmetries of maternal and paternal kinship. Ishmael was Abraham’s son by Hagar, an Egyptian slave. While Ishmael dwelled with Abraham’s extended family he would have been surrounded by many individuals with indirect copies of the genes he received from Abraham, but no individuals with indirect copies of the genes he received from Hagar, excepting Hagar herself. If Ishmael were to have performed some action that betrayed his father’s household but benefited himself, this action could well have had negative consequences for the genes he received from Abraham, because of costs to the patriarchal family, but positive consequences for the genes he received from Hagar, because of benefits for himself. If Ishmael faced such a choice and similar situations had been repeated many times during human ancestry, one might expect his maternal genes to add a little weight to the scales on the side of betrayal and his paternal genes to add a little weight on the side of faith in the patriarch.
Asymmetries of relatedness are greatest in an offspring’s relations to its parents, and it is here that the potential for internal genetic conflict is predicted to be strongest. A daughter’s maternal genes are definitely present in her mother, but the daughter’s paternal genes are absent from her mother. Therefore, the maternal genes of daughters value benefits to mothers as highly as benefits to self, but benefits to mothers are of little if any value for the paternal genes of daughters. (Here, I gloss over some complexities due to variation in mating systems.) Significantly, although internal genetic conflict is present in a daughter with respect to her relations with her mother, the same is not true of the mother’s relations with her daughter. The mother’s maternal and paternal genes have equal probability of being present in the daughter. Thus, a child’s feelings toward its parents are predicted to be more internally conflicted than the parent’s feelings toward the child.
Consideration of conflict between maternal and paternal genes would be mere sophistry if not for the fact that some genes exhibit behaviors that vary with the sex of their most recent origin. This phenomenon is known as genomic imprinting because it is assumed that some mark, an imprint, becomes associated with a gene in a parent’s gonad. This imprint must then be transmitted to the gene’s direct copies in the next generation, identifying the gene as coming from a mother or father, but the imprint must be able to be erased in the gonads of offspring, so that the gene’s copies can inherit the appropriate imprint in grand-offspring. An imprint is a contingent, rather than a fixed, property of a gene because a paternal gene in a daughter will be a maternal gene in the daughter’s son. Therefore, the paternal imprint must be erased and reinscribed as a maternal imprint. Imprinted genes are actively expressed when inherited from one sex, but silent when inherited from the other. Thus, a past environment, whether a gene resided in a male or female body in the previous generation, can influence the gene’s expression in the current generation.
Imprinted genes influence brain development and function. Barry Keverne and colleagues were able to produce mice with two different kinds of cells in their bodies (Keverne et al. 1996). Some cells possessed genes inherited from a mother and father, whereas other cells possessed genes inherited from only one sex. Cells that lacked maternal genes were well represented in the hypothalamus of the mouse brain but absent from the neocortex, whereas cells that lacked paternal genes were well represented in the neocortex but absent from the hypothalamus. These observations suggest that maternal and paternal genes perform different roles during normal development of the mouse brain, and hint that paternal genes favor relatively greater weight for hypothalamic preferences in decision making whereas maternal genes favor relatively greater weight for neocortical preferences.
As a gross oversimplification, the hypothalamus controls “visceral,” and the neocortex “cerebral,” motivations. Among primates, the relative sizes of neocortex and hypothalamus correlate with the composition of social groups; in particular, neocortical size increases with the number of adult females in a typical group (Keverne et al. 1996). Two asymmetries of mammalian social groups may explain a bias toward genes of maternal origin motivating “other-directed” neocortical behaviors and genes of paternal origin motivating “self-directed” hypothalamic behaviors. First, uncertainty of paternity and predominant maternal care mean that social ties will often be stronger among maternal half-siblings than paternal half-siblings. Second, male-biased dispersal at reproductive maturity means that most mammalian social groups are based on matrilineal kinship (Haig 2000a), although humans may be a partial exception (Haig 2010b, 2011b).
For current purposes, I merely want you to entertain the possibility that maternal and paternal genes have conflicting interests, so that I can ask how such conflicts might be expressed within the mind. Imprinted genes would be expected to influence broad behavioral tendencies and personality traits, rather than micromanaging every individual decision. As Ishmael is tempted to betray his father, his genes cannot be well-informed about the details of his particular dilemma. Rather, his genes would have instinctual information about outcomes of similar choices in the past when more or less weight had been given to the recommendations of temptation versus conscience. If, for example, maternal genes had benefited, on average, when relatively more weight was given to temptation, then maternal genes would be expected to evolve adaptations to promote a greater tendency to succumb to temptation. But these adaptations of maternal genes would be opposed by adaptations of paternal genes to increase the suasive power of conscience. This internal tension could be played out during development, with maternal and paternal genes favoring growth of different brain structures, or it could be played out during brain function, with one set of genes enhancing, the other dampening the amplitude of particular neural signals.
What are the internal factors that determine our choices? Clearly, our natures are strongly influenced by the unique set of genes we inherit from our parents, but they are also influenced by the beliefs and memories we have accumulated during our lives. Our choices are influenced by ideas just as much as by genes. Most choices are simple, but we sometimes need to choose in situations when genes and ideas provide no clear guidance or when different internal voices offer different advice. Conflicts can exist among ideas, among genes, and between genes and ideas. The self can be viewed as the entity that is held responsible for our choices. We are free actors at least in the limited sense that no single set of interests exclusively determines our choices. We are also free in the sense that no one, not even ourselves, can predict with complete accuracy how we will choose in all situations.