10
Social Intelligence and
Culture

Social Environment
It is critical for the biological success of many species.
It is cognitively complex.
Most social animals’ awareness of their social environment consists on
how others relate to themselves (mother, sibling, …). They’re not aware
on how other relate to others.
Primates are sensitive to third parties relationships (e.g.: they redirect
aggression against members of the family of those they have been in
conflict with).
Hence, successful interaction among these primates require a system of
proto-beliefs. This is also suggested by the fact that (i) information is
often not acquired at the time it is used and (ii) that information can be
relevant to more than one action.

The Social Intelligence Hypothesis (SIH)
Females-choice among primates often rests on the performance done
by a male in his past (e.g. he provided resources to his partner). Since
this information is not available on line it is likely that primates
construct social maps (e.g. maps of family relationships and social
hierarchies).
This is a plausible origin of belief-like representations. These
representation are reliable insofar as family relations are rather stable
and don’t change drastically over time. This is evidence for SIH.
Memory demands: the recognition of other demands recognitional
capacities and memory. For, an agent’s own status depends not only on
sheer size and strength but also on her social network.
Social learning: primates manifests some form of social learning.
Different animals in their group have overlapping information about
their environment (cf. Dennett “information gradient”).

Social Intelligence
A social map is an important tool for a social animal, for it allows
representation not tied to specific responses but also of possible
actions.
The social intelligence hypothesis is not simply a good explanation of
the evolution of the cognitive architecture pictured by folk psychology.
It is also a good explanation of the emergence of the interpretative
capacities themselves. For the demand of social life forces the great
apes to interpret other agents by tracking their psychological states
(e.g. detect cheaters).
The increasing demand of social life drives the evolution of agents with
intentional psychology for they need: (i) rich representations, (ii)
decoupled representations, and (iii) control of actions.

Great Apes and Imitation
They may have been capable of tracking other agents mental states
within their social group without having a theory of mind or mindreading capacity. They cognitive skill may not depend centrally on a
psychological model of other agents. It may rest on their learning by
imitation.
Imitation is not necessarily a theory of mind task, but it is a cognitively
sophisticated one. To successfully interact in a social world a
chimp/great ape need not have developed believes about believes as
the social intelligence hypothesis suggests.
A behavior program enabling them to learn by imitation may suffice.
Thus a primate/great ape storing proto-information of the sort:
If A hears a noise like so, A will act like this
where ‘so’ is a sensory specification of the noise and ‘this’ is a specific
motor pattern.
E.g.: Bonobos appease angry behavior by trading sex (see Waal).
Two explanations: (i) social intelligence hypothesis claims that they
track psychological states (angry); (ii) it may be that anger-mediated
behavior have a distinctive sensory signature and that bonobos merely
responds to it. They thus track the sensory cue, not the psychological
state.
Experiments (Piovinelli) show that chimps don’t track attention: they
have only rudimentary capacities for registering attention. Hence if
they are incapable of registering attention they are a fortiori unable to
read other chimps minds.
Chimps also have difficulties tracking/understanding pointing.
The chimps understanding of attention depends on automatic and
quasi-perceptual mechanisms.

Moral
The basic idea of social intelligence hypothesis is right. Social
interaction is strategic and the fact that there is a feedback loop build
into the picture of cognitive evolution makes it easy to understand that
social intelligence was an important element of primate cognitive
evolution and that social life was an important driver of the evolution
of belief-like states in our lineage.
Human social complexity, though, much exceeds that of great apes. So
if apes social life does not require them to have a theory of mind, this
may not be true of humans social life.
These considerations do some work to bridge the gap between social
and psychological competence.

Behavioral abilities vs. intentional agents
Questions: (i) what distinguish a habit machine from an intentional
agent?, (ii) what selective advantages do these abilities confer?
Intentional agents represent the goals of their own actions. Intentional
agents have a utility function, not just drive.
Intentional agents, unlike habit machines, can learn what is good for
them.

Reconstructing Hominid Evolution
Methodology (Cosmides & Toby):
1. Consider the problem our ancestors would have needed to solve,
given their way of life and environment (i.e. foraging and hunter
gathering in the Pleistocene)
2. Develop a theory of the cognitive mechanisms needed to solve those
problems
3. Specify the ways such mechanisms would be manifested in
development and behavior
4. Once (3) is complete, deploy the experimental techniques of
developmental, cognitive, and social psychology to test for those
mechanism’s presence. If they are discovered that confirm the scenario
of (1) and (2).

Cosmides & Toby:
We evolved in social environment characterized by a good deal of
social cooperation with the capacity to detect cheaters.
We should reason well about social exchanges and we do.
We have an adaptive specialization for social exchanges and detection
of cheaters.
Cooperation and group selection/development, as we saw, must have
plaid a crucial role.
Cognitive and social innovations that support a system of rewards and
punishment form the key to the human cooperation explosion (Sober &
Wilson).

The Modularity Hypothesis
Evolutionary psychology presupposes that our mind cognitive
capacities rests on an ensemble of specific-purpose computational
mechanisms. I.e. we come pre-equipped with cognitive modules
adapted to solve specific problems pertaining to our biological, social,
and physical environment.
Main example is language (Chomsky, Fodor, Pinker, etc.).
Yet the function making modular theories of language plausible are
distinctive and not shared by other domains. Thus massively modular
theories of mind should not be based on the example of language.
Chomsky thought that we don’t learn language via a general-purpose
learning mechanism. We don’t learn language, language instinct,
language organ, poverty of the stimulus,….

Syntax and Intentions
From an evolutionary perspective language developed as a mean of
transmitting beliefs/decoding information. At this stage syntactic
interpretation is crucial.
In a second stage it also developed as a mean of ordering, cheating,
etc. At his stage people must interpret what one says, not merely
decode the message.
We thus must distinguish understanding from acceptance.
The language module plays an important role in the interpretation but
the view cannot be generalized since at this level of interpretation
there is no treat of deception. This level of interpretation merely
depends on the independence of the organizational feature of
language from its content.

Hybrid modularity (Fodor)
Some decision problems, unlike language and perception, are not
suitable for a modular solution. Some problems are informationally
bounded while other are informationally open. (e.g. jealousy).
We have hybrid mental architectures.
We have in common a set of specialized cognitive capacities. These
explain our fast, autonomous and fairly uniform performance on input
tasks.
We also have central processors whose ontogeny and operation is very
different from that of a module.
Our modules have epistemic limits while we don’t.

Domain specificity
Evolutionary psychologists (Cosmides & Toby) take it to be the defining
criterion of a module.
A domain corresponds to related sets of adaptive problems (which the
agent must solve to survive and reproduce) environment pose for
agents. (e.g. of defined domains: foraging, evading predators, choosing
mates, …)
Modules evolved as a reflection of the specific information-processing
problems defined by these domains.
Modules are also adaptive specializations.
General purpose capacities play a relatively minor role in explaining the
distinctive features of human cognition (Sperber denies the existence of
a central processor at all).

Massive modularity
If the massive modularity view is right we should expect (i) cognitive
variations across different problems and (ii) relatively muted variations
across human groups.
1. Variability. When human faces problems for which they lack
biological preparation they should show cognitive blindness. The
commitment to innate module goes with the commitment to the
existence of cognitive limits.
We can manage money and bank account, for instance, even if they are
recent phenomenon because modules can be sensitive to input for
which they have not been specifically designed (e.g.: our resource
assessment module may recognize money as a resource).
At the right level of analysis, cross-cultural commonalities emerge (e.g.:
religious practices).

Argument for Massive Modularity (Sperber)
Loosening the domain of a module does not create greater flexibility
but great danger in the organism’s response to a problem (e.g.: if an
organism has a module detector avoiding being smashed by elephants
it is unlikely that this module will be as operative if it generalizes, i.e. a
mutation in the module would create greater danger for the organism).
Sperber suggests that radical changes in the environment may have
generated quite a dramatic disjunction between the domain for which
a device is currently used and the reasoning task which explains its
evolution (the elephants module detector may end up detecting rhinos
or trains). The module gradually acquires a new function.
Sperber’s tale takes place in an unrealistic transparent environment. If
we take on board noises and errors, it may be that the elephants
detector generalizes as a less specified module, i.e. can develop to
greater or less level of encapsulation (Sperber’s original module
detects both elephants and rhinos, but it becomes more suited to
detect rhinos than elephants).

The poverty of the stimulus argument, PSA
It is probably the main argument favoring the massive modularity
hypothesis.
Words could not be learned from the information available to children
by general-purpose learning systems.
The ontogeny of the competence is relatively independent of exposure
to information: various information don’t generate variations in
development (we have cross-cultural uniformity).
Since it is hard to develop from experience the right concept to
describe the data, language is likely to be an eccentric stimulus domain
(Fodor) because the salience features of utterances are unlikely to be
evident to our general perceptual and cognitive mechanism.

Two interpretations of the PSA.
1. Nativism. The information deficit is made up by information build
into the mind.
2. Perceptual specializations. The information deficit is made up by
special perceptual tuning to salient features of that domain.
Hence we have two evolutionary responses, i.e. cognitive development
or perceptual development.