Parte II.
Ensayos sobre la cognición social y el lenguaje
Sección I.
Ensayos sobre la evolución de la cognición social
The nonlinear evolution of human cognition
H. Clark Barrett
University of California - Los Angeles (UCLA)
Humans are special. We tell ourselves this all the time, and maybe we are right. Over the last several hundred thousand years we have populated the globe, mastering both our physical environments and our social ones. It is widely held that these achievements have something to do with how smart we are, and in particular, with our social intelligence. Compared to other primates, humans appear to be social geniuses: we are able to live peacefully, and even cooperate, in groups both large and small. We are experts at reading each others’ thoughts, feelings, intentions, and desires. We transmit information using elaborate systems of symbols, leveraging the accumulated knowledge of those who have gone before us to transform our landscapes, domesticate our food, govern ourselves, and land on the moon.
All of this cries out for explanation, and plenty of explanations have been offered. Just as scientists in the early twentieth century searched for the «missing link» that separated humans from other apes, scientists in the twenty-first century have searched for the evolutionary «it factor» —the psychological skill or capacity that separates human cognition from that of our fellow primates. Is it social learning? Symbols? Concepts? Language? Theory of mind? Prosociality? Morality? Neural plasticity? Bigger brains? Insert your favorite candidate here.
All of these, of course, are things that humans have, make, or do —that is not in question. What is hotly debated in today’s evolutionary literature is the question of which of these factors —or more likely, which combination of them— explains human psychological uniqueness. And since all evolutionists are materialists who accept that our minds are made of physical stuff —neurons, neurotransmitters, genes, and other biochemical ingredients— this question obliges us to ask what changes in brains have occurred in our evolutionary lineage to give rise to the new abilities that humans have. And here is where, despite decades if not centuries of accumulating data, there still seems to be no consensus in sight.
Why not? Part of the answer, to be sure, has to do with missing pieces of the puzzle: although we’ve accumulated lots of data about the brains and minds of humans and other primates, we don’t yet have enough for a definitive and complete answer to the question of what makes them different. But a glance around the scholarly landscape suggests that this isn’t all: our problems are also, almost certainly, theoretical. In this essay I would like to briefly consider one such theoretical problem: the problem of how to theorize psychological adaptations, and how they evolve in a mind composed of multiple, interacting adaptations. First I will describe what I think the problem is. Then I will explain how I think it is standing in the way of progress in understanding the evolution of the mind. Finally, I will close with some thoughts about how this problem can be solved.
1. The problem
In 1983, philosopher Jerry Fodor published his landmark book The Modularity of Mind, in which he formalized the concept of a psychological «module»: a sort of cognitive «reflex», in his terms, which he contrasted with what he called «central» (flexible and domain-general) systems of thought. In some ways, Fodor’s book revolutionized the field of cognitive psychology by helping to make explicit what is meant by a mental mechanism, and how we would go about studying one empirically. But in another way, the book has had a damaging consequence on the field by promulgating the widespread view that the mind is composed of two fundamentally different kinds of mechanisms: specialized mechanisms, or «modules», and unspecialized «central» systems. This is the view that I will call «psychological dualism» (Barrett, 2012). I will contrast it with what one might call a psychological diversity view: that the mind is composed of diverse adaptations, which in turn have diverse functions and take diverse forms, not just two. Importantly, these diverse forms can include properties that are thought to be characteristic of «central» systems —such as flexibility and interactivity— despite being specialized to carry out particular functions (Barrett & Kurzban, 2006).
Fodor originally proposed a set of nine features which he held to be characteristic of modules, and which distinguished them from central, non-modular systems: these included domain specificity, mandatory operation (automaticity), limited central access, speed, informational encapsulation, shallow outputs, fixed neural architecture, characteristic and specific breakdown patterns, and characteristic pace and sequencing of ontogeny (1983). At the time, he was clear that he didn’t intend modularity to be a binary property but a «matter of degree». In the literature, however, this subtlety has been lost, and modules are now seen as uniformly possessing this Fodorian package of traits. They are, in essence, like bits of psychological Lego that are rigid, reflex-like, automatic, innate, and independently operating. Non-modular systems, in contrast, have the opposite set of properties: they are flexible, interactive, and shaped by experience. This contrasting set of features defines a dualist taxonomy of mental mechanisms. In the «dual systems» literature, these are sometimes called System 1 (modular) and System 2 (non-modular), and System 1 is often explicitly associated with «specialized» psychological adaptations (Chiappe & Gardner, 2012; Kahneman, 2011; Stanovich, 2004).
There are many problems with psychological dualism, about which I have written in detail elsewhere (Barrett, 2005, 2006, 2012; Barrett & Kurzban, 2006, 2012). Here I would like to focus on one set of properties of putative «modules» which, if it were truly characteristic of psychological adaptations in general, would have extremely important implications for the study of mental evolution, and in particular, for at least some conceptualizations of the «it factor». I will refer to this set of properties as autonomy. Other suitable words might include independence and isolability. There are several ways in which «modules» or «adaptations» are widely held to be autonomous, independent, or isolable, all of which are probably, in a strict sense, wrong.
2. The non-autonomy of mental evolution
What I mean by autonomy is a mix of two related components —architectural autonomy, implying that modules can and do operate independently of other systems and can be removed while leaving the operation of other systems intact— and evolutionary autonomy, implying that modules are shaped by evolution independently of other systems, such that their functions are relatively isolable.
In the evolutionary developmental literature, modularity is often seen as related to the question of «evolvability»: the degree to which evolutionary forces, including mutation, drift, and natural selection, can independently shape different aspects of organismal phenotype, as a function the «decomposability» of the phenotype into relatively independent parts or subunits (Wagner & Altenberg, 1996). Modeling work, as well as empirical studies of the modularity of existing development systems, has shown that the answers are neither obvious nor entirely intuitive (Wagner and others, 2007). Importantly, it is likely that no organismic traits are entirely autonomous, either architecturally or evolutionarily. Degrees and aspects of autonomy must and do play a role in shaping the course of evolution —but so do degrees and aspects of interactivity and non-independence. And perhaps most importantly for thinking about the mind, this doesn’t mean that everything inside the organism is one big mishmash: there is functional specialization and modularity, but the important evolutionary questions lie just as much if not more in how the specializations interact, as in their relative autonomy or independence.
There are several basic facts about how evolution occurs that make strict autonomy a poor diagnostic for psychological adaptations. The first is that new adaptations evolve from old ones. Therefore, «new» psychological design inherits much «old» design. This means that multiple psychological mechanisms might, indeed must, share aspects of their information-processing design because they inherited it from older mechanisms they both evolved from (Barrett, 2012). As a consequence, these mechanisms might not appear «autonomous» along dimensions of shared design, and indeed, would be empirically impossible to disentangle if investigated solely along those dimensions.
For example, it has been proposed that human minds might contain specialized subsystems for recognizing and processing different kinds of physical objects, including faces, artifacts, and places (Kanwisher, 2010). Critics of this proposal point to aspects of object processing that can be shared across object domains, such as «configural» processing, as evidence against such subspecialization (Gauthier & Nelson, 2001). But if these subsystems evolved from a shared common ancestor and are now to some degree phenotypic...