But
there are many critics of EP's view of the mind, in particular the highly
problematic notion of modules, or self-enclosed and informationally encapsulated
organic computing devices. As I hope to show in what follows, the modular
view of mind (MM) is simply not adequate to describe the complexities of human
thought, and it rests upon certain assumptions that are themselves questionable.
"With Friends Like These. . . ": Jerry
Fodor on intentional fallacies
One prominent (and surprising) critic of the modular
theory of mind in the past few years is one of its founding fathers, Jerry
Fodor. In a recent book he points out several large problems created by assuming
MM, especially for global kinds of mental processing. One difficulty is the
"frame problem": despite some advances in understanding how the
brain works, we are light years from being able to produce a computer that
can do the simplest tasks we often entrust to a human child, such as ".
. . making breakfast without burning down the house." But Fodor notices
the over confident tone in Pinker's book: "It's a striking peculiarity
of [his] book in particular that he starts by remarking how hopelessly
far we are from being able to build a serviceable robot, but never explains
how to reconcile our inability to do so with his thesis that we know, more
or less, how the cognitive mind works" (37). This is particularly strange
because Pinker bases so much of his analysis of the human mind on the computational
theory of mind (81-148)
While Fodor is willing to grant that some of our lower cognitive functions
may indeed be modular, he worries about what he calls "massive modularity",
the view that posits modules for just about everything we think the mind capable
of doing. One key question, for example, is just how a module "knows"
when to activate itself, how it decides that a given piece of input data is
relevant to its sphere of functional sovereignty:
Suppose
that Darwinian processes have somehow endowed homo sapiens with an encapsulated
[cheater detection module] that deploys domain-specific inferential procedures.
. . to evaluate what are taken to be social exchanges. Even so, that wouldn't
be a reason for thinking that the mind is really massively modular, unless
you are also prepared to suppose that corresponding encapsulated and domain-specific
procedures could evaluate situations at large for whether or not they are
social exchanges. But. . . figuring out whether something is a social exchange
and, if it is, whether it's the kind of social exchange in which questions
of cheating can arise, takes thinking. Indeed, as far as anybody knows, it
takes the kind of abductive reasoning that, by definition, modules don't do
and that (it appears) Classical computations have no way to model. (77)
Fodor raises another objection when dealing with MM attempts to explain
human cognition by appeals to adaptationist "just-so" stories. Along
with the obvious difficulties inherent in such a project, such as the inability
to do empirical work on humans' original environments and their inter-actions
with it, there is also the problem of teleology, or what something "was designed for by natural
selection".
According to Fodor, a normal assumption of EP theorists such as Pinker and
others who hold to the MM thesis is that, a priori, most explanations of mental
activity must fit with a Darwinian adaptationist framework when this may often
need to proven before it can be assumed.
As Fodor states, "So if you want a theory of cognitive architecture,
you need a notion of function; and if you want a notion of function, you have
to be an adaptationist about cognition. So the argument goes" (85). But
this begs the question under discussion.
Further, is it really true that we must have a Darwinian, diachronic explanation
for why a particular organ (in this case, the brain) does what it does? Couldn't
we just as easily make synchronic discoveries, or offer synchronic explanations, which,
ipso facto, preclude knowing an organ's selectional history? Even if the notion
of Darwinian function is correct, it still doesn't follow that diachronic
methods are necessary. We might just as easily determine the function of hands
are for holding or birds' wings are for flying by simple observation without knowing their
evolutionary history (86):
It is (yet another) species of intentional fallacy
to argue that if A=B, then you can't know about (figure out, have a theory
of, explain, make justified claims about. . .) A unless you know about (figure
out. . .) B. . . . [N]on-Darwinian notions of function can easily be imagined.
My intuition, for example, is that my heart's function has less to do with
its evolutionary origins than with the current truth of such counterfactuals
as that if it were to stop pumping my blood, I'd be dead. Maybe it's the case
in general that what determines an organ's function is something about which
such counterfactuals are true of it. (86-87)
Would
a better explanation, one that includes Flanagan's accounts of exaptations
discussed earlier, include both synchronic and diachronic elements (at least,
where the latter can be more reliably inferred)?
Further,
some EP proponents believe that the fields of biology and psychology must
be related, consilient, and/or mutually interdependent, and this must be done
via Darwinian synthesis. But Fodor rightly points out (80-84) that most valuable
scientific work often happens within a specific discipline by scientists who
are working on solving a particular problem and do so by discovering a particular
causal mechanism or relevant explanation for the phenomena in question-- and
for many discoveries, as valuable as they are within their discipline, there
the significance ends. There is little, if any, overlap with other fields.
While it is trivially true that all contingent truths are consistent with
each other (i.e., "Lions are carnivorous" is consistent with
"Most schizophrenics have a psychotic disorder"), it does not necessarily
follow that mammalian eating habits have much to do with human cognitive functions,
and vice-versa. But EP advocates often hav ea vested interest in using MMdespite
its limitations:
Holism
Revisited: The attack of developmental systems theorists
Others
critics have attacked the view of brain function held by SB and EP researchers
on different grounds. One such is Barbara Herrnstein Smith, who points out
two central assertions EP advocates commonly make:
The
first is the idea that the mind, like a computer, is an information-processing
machine, the purpose of which is to solve problems through rule-governed manipulations
of symbols that represent objective features of the world. The second is the
idea that the mind, like the body, consists of multiple individual organs
or "modules" engineered by natural selection to maximize the reproductive
fitness of our Upper Paleolithic ancestors and reflecting that design more
or less directly in their current operations. (158)
But
as she makes clear (155-172), there are numerous problems with this model
of the mind. Along with recurrent equivocation of basic terminology like "mind"
and "brain" (which can mean, variously, "mental organs", "neural circuitry",
and "our thoughts and feelings"), this methodology assumes an epistemology
that relies on a world that is static, objective, external, and unchanging.
As
one example, Smith uses Pinker's description of the problem our mind has in
obtaining information via visual perception, an obvious need for survival.
From this Pinker infers the existence of neural programming designed by natural
selection to solve just such a dilemma-- but this reasoning is circular: only
a certain view of the world would lead to the creation of a "problem"
that could then be solved by the view itself:
For it is only in accord with the realist/representationalist
assumptions of that model that visual perception would be seen as defective
informational input in the first place, or that the brain would be seen as
requiring "correct representations" of the world in order to direct
behavior appropriately. Other current models of cognition that describe visual
(and other) perception without such ontological and epistemological presuppositions
can also offer accounts of appropriate (that is, self-sustaining and effective)
organism-environment interactions without the postulation or "deduction"
of extensive preloaded computational software. (158-59)
One need not (and indeed, I do not) necessarily agree with Smith's somewhat
hasty dismissal of realism to see the value of her analysis here. Indeed,
she has illustrated a common tendency in much work done by various EP theorists,
that of assuming at the outset certain metaphysical and epistemological presuppositions
that need to be carefully defended before problems are confronted and explanations given. Sloppy
philosophy may prevent the formulation and confirmation of good scientific
theories-- or, at the very least,
we should expect EP advocates to defend these often-unexamined antecedent
beliefs.
Smith also faults EP theory for not taking into consideration that human
beings (as well as other biological organisms) are living systems which interact
in complex ways with their environment over time (Lewontin, 2000; Oyama, 2000
)-- humans are not simply hardwired computer-like brains operating in a totally
rule-bound manner. Any analysis, therefore, of human systems will need to
recognize this:
Major
alternative theoretical frameworks for explaining intelligent
human behavior and cognitive processes more generally give due weight to the
relevant characteristics of living systems. Rather than a series of discrete
problem-solving computations, the processes of cognition in human beings--
as in organisms more generally-- may be seen as the continuous modifications
of our structures and manners of functioning in the course of our interactions
with an always, to some extent, changing environment. Proponents of such alternative
ecological and/or dynamical models of cognition stress that humanlike
intelligent behavior includes embodied skills that cannot be translated into
symbolic manipulations or statementlike propositions, and that perception
is an interactive activity, not a form of spectation or the passive reception
of unidirectional environmental "inputs."
This
kind of program, Smith rightly points out, will emphasize the important role
of cultural influences, education, and various institutions such as law and
religion in shaping human development, both individually and corporately (167-68).
Further,
Smith believes we must carefully draw a distinction between behaviors that
indeed appear innate, such as the startle-response, and
behaviors which may develop more slowly, during a particular phase
of an organism's development, such
as stereoscopic vision. This means that something that might regularly develop in a human child could be derailed
by outside influences such as those listed above. But EP does not always develop
these more sophisticated categories, so its theoretical apparatus remains
fairly simple, leading to overly simplistic explanations of what are in fact
far more intricate developments.
Further
evidence against the MM position often adopted by EP proponents comes from
studies carried out on infants, children, and adults with Williams Syndrome
(WS), who are able to achieve proficiency in facial recognition, speech, and
social relations, in sharp contrast to their severe impairments in spatial
cognition, number and problem solving. The dissociations between impaired
and seemingly preserved domains in WS are typical of those also described
for brain-damaged adults. But do such findings necessarily imply domain-specific
beginnings in which some modules are preserved and others impaired? (Karmiloff-Smith,
178)
At
first glance, WS appears to confirm the modular thesis, but as Karmiloff-
Smith points out, "if we could demonstrate that people with WS solve
language and face-processing tasks by processes that are different from those
used in normal development, then their relatively intact behavior cannot be used to make evolutionary claims about the
pre specification of domain-specific modules" (179).
Indeed,
what Karmiloff-Smith and her associates discovered was that people with WS
used different, learned procedures to master one of the skills in question,
facial recognition:The excellent scores on the face recognition and memory
tasks highlight the fact that even though behaviorally people with WS seem to be a prime example of the sparing of a face-processing
module fashioned by evolution, and on which their retardation in spatial cognition
has no effect, it turns out that they are in fact successfully solving the
face-processing tasks via a different strategy. Thus their successful behavior
cannot be invoked to argue in favor of a preserved, innately specified face-processing
module, even if such a module might normally exist. Empirical research on
WS gives no support to the [EP] claim. (180)
They
discovered similar findings when investigating WS patients' language acquisition
skills as well as normal and atypical infant development (180-83).
The
evidence is strong in these cases that much, though not all, of human development,
including cortical growth, is regulative, not
just mosaic: that is, the ontogeny of the human cortex i s "under broad
genetic control; it is slow and progressive, and numerous parts of the system
develop interdependently. Like mosaic development, it is vulnerable under
non optimal conditions, but.... it is flexible in the face of change…[1] n
processing the environmental input its resulting complexity is much greater
than mosaic development" (184).
In
other words, the analogy of an infant's brain as a perfectly formed Swiss
army knife is misleading-- we would be better to direct our attention to
explaining the development of the knife:
"Thus we should explore how each of the specialized parts of the army
knife develops over time from a series of simpler tools. That is, the knife's
ontogenesis" (185). Further complications for the EP theory of mind,
based on mental modules and Darwinian algorithms, develop when we consider
what Sterelny and Griffiths (1999:329-331) refer to as poverty of
the stimulus arguments, which ". . attempt to show that we
develop cognitive skills too fancy, and with too little information from the
environment, for their development to be the result of general learning mechanisms."
Examples here include visual system outputs, which take fragmentary inputs
and connect them to produce a complex end result, and mating selection:
The
. . . mating rule is not computationally complex. It is very simple: women
find high status attractive, whereas men find youth attractive. There is no
need for a specialized mechanism to operate this decision rule. It may, of
course, be very difficult to determine whether someone is young or of high-status,
but that is not what the specialized mechanism has to do. There are many social
interactions in which age and status judgments are important, and there is
no evidence that our judgments of age or status for the purposes of mating
ever conflict with the judgments we make for other purposes. So there seems
no reason to suppose that assessing age and status is the work of the postulated
mate choice module. All the module does is direct our sexual attention to
those who have these properties, and that is a relatively simple task. (330)
Further, one simple mechanism does not appear to be complex enough, on the
other hand, to handle the vagaries of human language. While we may be able
to interpret the literal meaning of a sentence, verbal communication is often
far more subtle than this, involving such things as intentionality and whether
we trust the speaker: "[These problems are] not solvable by shortcuts
from a restricted data base-- that is, by an encapsulated device. . . Could
an encapsulated mechanism deliver reliable judgments about a prospective mate's
status? A spouse's infidelity?. . . We need to be shown the equivalents in
these domains of the reliable rules of thumb that our perceptual mechanisms
exploit" (331). Also, since there is good reason to believe that organisms
with cognition actually change their environments, any explanation that starts
with a problem to be solved and looks backward for environmental clues will
be in serious jeopardy of deriving dubious confirmation of their hypothesis
(332).
Finally,
it seems rather odd that emotions play such a small role in the EP conception
of mental functioning. While no one would claim that emotions can alter the
laws of logic or even that human beings should primarily be guided by emotional
states divorced from reason, emotions nevertheless appear somewhat neglected
by modular theorists.
This
is even more surprising when we consider that certain basic emotions such
as anger and surprise may indeed be modular (Griffiths, 90-98), similar to
lower-level cognitive functions which Fodor and others d o think could be
modular (though Fodor's criticisms of EP's tendency towards Darwinian adaptationism,
diachronic explanations, and false teleology need to be applied as well to
Griffiths where applicable).
Steven Rose (316) uses an example provided by Pinker: Robinson Crusoe finds
a footprint in the sand on his island, and this reveals information to him.
But Rose goes a step further, asking what kind of meaning Crusoe might be inclined to give to this information after it is recognized
and deciphered: Is he surprised? Afraid that the footprint's owner may be
hostile? Overjoyed at having a fellow human being to help end his loneliness
and isolation?
These natural questions illustrate " [a] turmoil of thoughts and emotions
within which the visual information conveyed by the footprint is embedded.
The key here is emotion, for the key feature which distinguishes brains/minds
from computers is their/our capacity to experience emotion" (316). The
MM approach to cognition pays scant attention to the role of emotions in helping
shape our response to the world around us. According to developmental theorists,
this response is part of the interaction all organisms have with their environment,
an interaction that shapes not just how we feel, but what we think about both our emotions and other reactions we have
to our environment.
The End. . . ?
As we have seen, the use to which EP proponents have put the modular theory
of mind is not a coherent one. Too many conceptual and logical problems are
left unresolved, too many other important questions are
ignored, to make it a compelling theory of
higher cognitive functions. While it may be unrealistic for us to expect more
modesty in both attitudes and prediction from people such as Stephen Pinker
and others, they would be
wise
to heed Fodor's words: "Hubris in cognitive science is particularly to
be avoided since it is not merely impertinent but also inaccurate. In fact,
what our cognitive science has done so far is mostly to throw some light on
how much dark there is. So far, what our cognitive science has found out about
the mind is mostly that we don't know how it works" (100).
Bibliography
Churchland, Paul M. The Engine of Reason, The Seat of the Soul: A Philosophical
Journey into the Brain. (Cambridge, MA: MIT Press, 1995)
Crick, Francis. The Astonishing Hypothesis: The Scientific Search
for the Soul.
(New York: Scribner, 1994)
Dennett, Daniel. The Intentional Stance. (Cambridge, MA: MIT Press, 1987)
Flanagan, Owen. The Science of the Mind. (Cambridge, MA: MIT Press, 1991)
Fodor, Jerry. The Mind Doesn't Work That Way: The Scope and Limits of Computational
Psychology. (Cambridge, Mass.: MIT Press, 2000)
Griffiths, Paul E. What Emotions Really Are: The Problem of Psychological
Categories. (Chicago: University of Chicago Press, 1997).
Karmiloff-Smith, Annette. "Why Babies' Brains
Are Not Swiss-Army Knives," in Rose, Hilary and Steven, eds., Alas, Poor Darwin: Arguments
Against Evolutionary Psychology. (New York: Harmony Books, 2000)
Lewontin, Richard. The Triple Helix: Gene, Organism, and Environment.
(Cambridge,
Mass.: Harvard University Press, 2000)
Oyama, Susan. Evolution's Eye: A Systems View of the Biology-Culture Divide.
(Durham,
NC: Duke University Press, 2000)
Pinker, Stephen. How the Mind Works. (New York: W. W. Norton, 1997)
Rose, Steven. "Escaping Evolutionary Psychology,"
in Rose, Hilary and Steven, eds., Alas, Poor Darwin: Arguments Against
Evolutionary Psychology. (New York: Harmony Books, 2000)
Smith, Barbara Herrnstein. "Sewing Up the Mind: The
Claims of Evolutionary Psychology," in Rose, Hilary and Steven, Alas, Poor Darwin: Arguments
Against Evolutionary Psychology. (New York: Harmony Books, 2000)
Sterelny, Kim and Paul E. Griffiths. Sex and Death: An Introduction
to Philosophy of Biology. (Chicago: University of Chicago Press, 1999)
Wilson,
Edward 0. Consilience:
The Unity of Knowledge. (New York: Vintage Books, 1998)