Cognitive psychology, cognitive social psychology, neuropsychology, and anthropology have since the
early 1970s been dominated by the representational-computational view of mind (RCVM) or, more simply, representationalism,
a model defined by three premises: (1) People behave by virtue of their possessing knowledge, (2) Knowledge is constituted
by mental representations, (3) Cognitive activity consists in applying computational operations to these representations.
The author's aim is to "mark the limitations of RCVM and to propose a more balanced psychological picture,"
to show that "the representational-computational framework is inadequate in several fundamental respects and that it
cannot offer the basis for a general theory of human behaviour." (p. 2)
is sympathetic to other non-computational views of mind and does detail both the differences and possible relations between
his and other non-representational frameworks, his critique follows an independent line of attack, a thorough consideration
of psychological phenomenology (rather than computational, mathematical, and biological issues as connectionism has done).
He begins by asking: what are representations? And why are they considered necessary for cognitive theory? Shanon distinguishes
six different senses of "representations":
(1) the most general sense:
a representation is "something that stands for something else" and includes concrete, external kinds of representations,
as well as internal, mental ones;
(2) the experiential rationale, the simple, pre-theoretical
view: representations are phenomenological entities--i.e., part of our direct experience;
the naive rationale, also pre-theoretical: representations are the locus of mental activity--i.e., the way to account for
our mental activity, prior to knowing anything more specific about them;
(4) the epistemic rationale:
representations are a necessary substrate of meaning--i.e., the covert structures that make possible the fact that our expressions
of behavior (especially language) are meaningful and rule-like;
(5) the functionalist rationale:
representations are internal structures that carry out mediating functions needed to explain organismic behavior, which cannot
be fully accounted for in terms of environmental factors;
(6) most important to cognitive psychology
(and the principle target of Shanon's critique), the "technical-psychological," or semantic, rationale: representations
are well-formed structures of well-defined abstract symbolic entities constituting a complete and exhaustive canonical code
which is determinate and static, a "fixedness" that operates on three levels: (a) systemically, by the code's
being set prior to cognition, (b) semantically, by the code's specifying only one "representational characterization"
in any given state of awareness, and (c) temporally, as a result of the other two, by the code's characterizations tending
to be rigid and unchanging across time.
Shanon draws an important distinction: the
"vertical," philosophical perspective, on the one hand, says that representations are needed to explain the relationship
between overt behavior of conscious organisms and underlying factors that make behavior possible (be they the organism's
knowledge, the external world, or the biological substrate of the organism's body)--and, on the other hand, the "horizontal,"
cognitive perspective, says that representations operate within a single level, exemplifying relations such as sameness vs.
difference of meaning, entailment, presupposition, etc.
Cognitive scientists in linguistics, psychology,
and artificial intelligence research tend to take for granted the existence of the external world and how the mind succeeds
in relating to and establishing contact with it--sometimes going so far (see, for example, Fodor) as to say that it is impossible
for psychology to account for the connection of the mind to the world--focusing instead on relationships between the structures
and processes of the mind. This methodological solipsism places some basic psychological issues "outside the realm of
cognitive science," a result that Shanon deems unacceptable.
A major thrust of philosophical
critiques is to challenge the tendency toward reification--i.e., the claim that mental representations are entities; much
philosophical research since 1950 has questioned whether mental entities are real or even theoretically necessary. Cognitive
critiques such as Shanon's, on the other hand, concentrate on evaluating various paradigms of scientific research--i.e.,
determining whether a given theoretical framework is appropriate or useful in explaining psychological phenomena. Shanon argues
that semantic representations fail, epistemically, to explain the knowledge that people obviously have and, functionally,
to explain relationships between cognition and reality. In a third line of critique, Shanon shows that "fixed, static
representations" cannot account for the origin, development, and progression of knowledge. A fourth, secondary line addresses
more basic problems with the concept of "representation" itself.
Although Shanon places
more stock in the cumulative impact of his overall challenge to representationalism than in any of the specific arguments,
it is in Chapter 6, "The Unbridgeable Gap" and the five subsequent chapters that Shanon presents what to this reviewer
is his most compelling case against RCVM which, he says, "cannot account for relationships between...the cognitive system
and the body...the external physical world...the social order...and non-cognitive systems, primarily affect and motivation,"
nor between the body and the world. (p. 93) Nor, Shanon later shows, can RCVM "be salvaged by drawing a line between
knowledge-based behaviour and behaviour which is not based on knowledge, between the so-called cognitive and that which is
not." Not only would this move surely impoverish psychology by getting rid of various important non-"knowing-that"
kinds of knowledge (knowing-how, knowing others, ethical and aesthetic judgments), as well as "meaning and interpretation,
perception and action, motivation and emotion, consciousness and selfhood;" but it would be futile as well, for "knowledge-based
aspects of behaviour are neither well demarcated nor primary." Instead, Shanon says, "action in the world, not knowledge,
is the basis for both cognition and behaviour in general." (p. 238
If RCVM is wrong, as Shanon
has most surely shown, why is it still so widely accepted? The reason is several clusters of basic assumptions, which Shanon
boils down in Chapter 16, "Why Representationalism?," to a single substantive premise, which he attributes to Descartes:
dualism--and a methodological premise from Plato: purism. With dualism, cognitive psychology on the RCVM model accepts "a
sharp split between the internal and external domains," which "strictly confines psychology to the internal, mental
world." With purism, RCVM makes psychology over into a non-empirical, ultra-logical, rationalistic discipline. Together,
as Shanon convincingly argues, these mutually reinforcing assumptions "put cognition into a strait-jacket. Adopting them
is tantamount to wishing to do psychology without being psychological." (p. 257)
17 and 18, "A Picture of Mind" and "The Representational and the Presentational," Shanon turns at last
to his positive thesis. He asks: "What is the basic ability of the mind? What is the cognitive system essentially designed
to do?" RCVM, of course, says it is the manipulation of symbols, but Shanon says it is action in the world. (p. 268)
The key concept Shanon uses to elucidate his view is that of a "presentation," which he uses in both a generic and
a more specific sense. A presentation in general is any concrete, articulated expression, whether overt or mental. Those presentations
that are composed of well-defined elements along clearly differentiated dimensions Shanon says exhibit a "representational
profile," while those that are not well-defined he calls (following Langer) "presentational."
These profiles are the polar extremities of a representational-presentational continuum, which applies both to type
of expression and mode of interpretation, resulting in four paradigm cases: formal reasoning (when a representational expression,
R, is interpreted in a representational manner), art criticism (presentational, P, expression with R interpretation), discerning
non-verbalized meanings in conversation (R expression, P interpretation), and spontaneous thought, fantasy, speech, and action
(P expressions, P interpretation). Whereas RCVM limits cognitive science to the first psychological profile "and thus
misses much of the richness of human cognition," Shanon's model provides a way to portray the vast range of cognitive
phenomenology that results from the "on-going interplay between the representational and the presentational." (p.
Overall, Shanon's critique is comprehensive, internally coherent, and gets at the fundamental
problems with RCVM, in the process paving the way for his own integrated alternative view of mind. Yet, interestingly, he
also grants that the functions supposedly fulfilled by representations may actually justify postulating representations in
the naive, epistemic, or functionalist senses--pertaining to the need for a locus of mental activity, for an internal structure
in which knowledge is coded, or for functions that mediate the relationship between mind and world--but not in the stronger,
technical-psychological sense pertaining to semantic structures. Shanon claims that if these distinctions were kept in mind,
there would be no need to deny "the existence of cognitive structures and patterns of activity that exhibit a representational
profile." He argues instead that representations are not primary, neither conceptually, procedurally, nor developmentally:
"representational structures, if and when they exist, are the products of cognitive activity, not the basis for it."
(p. 18) Shanon thus wisely avoids throwing out the representational baby with the technical-psychological bathwater. What
more could one ask?
Perhaps a little more consistency and clarity from Shanon when he delves
into philosophical matters would be helpful. He characterizes the RCVM model as detached and culturally "objective."
(p. 293) Yet, he also conclusively demonstrates the bias and blatant subjectivism involved in RCVM's eschewing of relevant
facts that do not fit its dualist-purist framework. (332, 349) Granted, there are two distinct senses of "objective"
at work here, but it would not have been out of place to suggest that such obvious disrespect for facts descriptively and
normatively overrides the detached "scientific" attitude toward the constricted range of facts RCVM deigns to recognize
Shanon also implies that the only choice we have is representationalism's Platonic-Cartesian view of cognition
as internal and "in the head" or presentationalism's Aristotelian, external, "in the world" view of
cognition. (pp. 249, 268) And he says that representationalism is naive realism. (p. 243) Yet, historically, naive realism
has been equated with an external locus for cognition and often associated (rightly or not) with the Aristotelian/Thomist
tradition, as opposed to the more modern form of representative realism, usually connected with John Locke, that posits an
internal locus for cognition.
There is a third alternative, a more critical form of realism,
that places the locus of cognition in the relation between organism and reality, a view proposed by various thinkers
such as C.E.M. Joad, Roy Wood Sellars, Ayn Rand--and Shanon himself, as evidenced by his exhortation to cognitive scientists
to "look at the coupling of the internal and the external. It is only there that meaningful regularities are to be found."
(p. 269) It is the open-minded, scientific investigation of the many modes of interaction between an organism's cognitive
system and the other internal and external systems involved that Shanon so rightly champions against the narrowness and insularity
of the RCVM model. This, not Shanon's questionable philosophical and cultural observations, constitutes the lasting value
of his book.