Posner, et. al., on Foundations of Cognitive Science |
ABSTRACT In Foundatins of Cognitive Science the research issues center in cognition (attention, perception, memory), learning, cognitve processes, and abilities. The usual disciplines included and, as usual, communication studies is excluded. Adherence to the computational model of the mind is strong. The computational / connectionist controversy is muted. Looking ahead, the controversay seems unavoidable. The volume does leave the door open for communication studies in at least two ways: It has the requisite character of a conceptual doman to be investigated and it shares "content analysis" as a method. |
In his "Preface: Learning Cognitive Science," Michael I Posner writes: CS "deals with the nature of intelligence from the perspective of computation" (x). Since the volume is a collection of essays by well-known cognitive scientists, he may be regarded as their spokesperson but this point of view is his assessment. He places emphais on "symbols" and their "manipulation," which is consistent with the general view that CS concerns "representation" and "computation."
Posner views CS as an exploration into "many domains" that can be included in "a verified architechture of cognition" (xiii). Herbert Simon and Craig A. Kaplan in their overview of CS identify the "standard model" of "the architecture of intelligent systems" as "an information processing system" (7-10). Posner's comment about many domains reflects the view that the cognitive scientists take on a research area within the architecture of CS but do so from their respective disciplinary domains.
cognitive science research areas related to communication
The research issues covered in the volume are reflected in the titles of the essays in Part Two, "Domains." The topics covered in this section are:
- LEARNING
- language acquisition
- cognitive skill acquisition
- COGNITIVE PROCESSES
- mental models
- conceptualization
- induction
- problem solving
- COGNITIVE ABILITIES
- planing
- reading
- discourse
- action
- COGNITION
- attention
- vision
- memory
All of these areas are related to communication research but the substantial crossover in research interests are: learning, mental models, conceptualization, induction, problem solving, planning, discourse, action. But it is difficult to research discourse, for example, without considering the role of attention, reading, or memory. (See "comments" below.)
disciplines included and excluded
Taking the contributors as indicative of the disciplines that comprise CS, Posner includes:
psychology (PSY) |
Alan Allport |
| computation* (C) | Barbara J. Grosz John E. Laird Allen Newell Martha E. Pollack Zenon W. Pylyshyn |
| philosophy (PHIL) | Patricia Smith Churchland John Etchemendy Gilbert Harman |
linguistics (L) |
Jon Barwise Thomas Wasow |
| anthropology (A) | Roy G. D'Andrade |
| neuroscience (N) | E. Bizzi Ellen C. Hildreth Michael I. Jordan F. A. Mussa-Ivaldi Steven Pinker Terrence J. Sejnowski Shimon Ullman |
| systems (S) | Paul S. Rosenbloom Candace L. Sidner |
*computation is a general term that includes AI, cybernetics, computer science, or simply computatinal analysis. Terrence J. Sejnowski is from a department of biophysics but his contribution is in the domain of neuroscience.
Since persons from different disciplines collaborate, the following chart might be a more accurate index of the extent to which fields of inquiry associated with particular disciplines are studied in CS. The authors are listed by last name with abbreviations for their disciplinary homes. The column designated "area" lists the conceptual domain covered in the essay with abbreviations for the discipline with which it is associated. The entry "CS" indicates that the domain is specific to cognitive science.
| title | authors | area |
| Foundations of Cognitive Science | Simon (PSY), Kaplan (PSY) | overview (CS) |
| Computing in Cognitive Science | Pylyshyn (C) | computer science |
| Symbolic Architectures for Cognition | Newell (C), Rosenbloom (S), Laird (C) | computer science |
| The Architecture of Mind: A Connectionist Approach | Rumelhart (PSY)(N) | neuroscience |
| Grammatical Theory | Wasow (L) | linguistics |
| Model-Theoretic Semantics | Barwise (L), Etchemendy (PHIL) | linguistics |
| Experimental Methods in CS | Bower (PSY), Clapper (PSY) | CS |
| Brain & Cognition | Sejnowski (N), Smith Churchland (PHIL) | neuroscience |
| Language Acquisition | Pinker (N) | pyschology |
| Reading | Pollatsek (PSY), Rayner (PSY) | pyschology |
| Discourse | Grosz (C), Pollack (C), Sidner (S) | linguistics |
| Mental Models | Johnson-Laird (PSY) | pyschology |
| Concepts and Induction | Smith (PSY) | philosophy |
| Problem Solving & Cognitive Skill Acquisition | VanLehn (PSY) | pyschology |
| The Computational Study of Vision | Hildreth (N), Ullman (N) | pyschology |
| Visual Attention | Allport (PSY) | pyschology |
| Memory | Schacter (PSY) | pyschology |
| Action | Jordan (N), Rosenbaum (PSY) | philosophy |
| Geometrical & Mechanical Issues in Movement Planning & Control | Bizzi (N), Mussa-Ivaldi (N) | computer science |
| Cultural Cognition | D'Andrade (A) | CS (A) |
| Some Philosophical Issues in CS: Qualia, Intentionality, & the Mind-Body Problem | Harman (PHIL) | CS |
This chart reflects the "interdisciplinary" character of CS. Some research areas associated traditionally with a disciplinary field are studied under the auspices of CS by other disciplines. For example, action theory since Aristotle has traditionally been studied in philosophy. In Foundations of Cognitive Science, it is studied by a neruoscientist and a psychologist. Discourse is usually regarded a subject studied in linguistics. In this volume it is studied by a researchers from a "computational lab" (Grosz), an AI center (Pollack), and from an organization on systems (Sidner).
methods included and excluded.
The disciplines closest to Communication included in the volume are linguistics, anthropology, and philosophy. The emphasis on the computational research component in it is echoed in the linguistic entries. Gilbert Harman's "Some Philosphical Issues in CS" and Roy G. D'Andrade's "Cultural Cognition" leave the door open as their essays do not confirm conclusions with a computer simulation of the cognitive processes that they address. Moreover, D' Andrade's essay draws upon Rosch, Lakoff, Sweetser, and Langacker whose methods are not computational.
adherence to the computational model of the mind
As Posner indicates in his "Preface" the cognitive scientists in this volume are committed to the cmputational model of mind. In practice, this means that the researchers all "verify" their findings by replicating them with computer models. This includes neuroscientists contributing to the volume who employ parallel rather than serial processing.
the computational / connectionist controversy
In "The Architecture of Mind: A Connectionist Approach," speaking of the differences between serial and parallel computational processes, David E. Rumelhart writes:
How does the replacement of the computer metaphor with the brain metaphor as model of mind affect our thinking? This change in orientation leads us to a number of considerations that further inform and constrain our model-building efforts. ... From conventional programmable computers we are used to thinking of knowledge as being stored in the state of certain units in the system. In our systems we assume that only very short-term storage can occur in the states of units; long-term storage takes place in the connections among units. Indeed it is the connections—or perhaps rules for forming them through exerience—that primarily differentiate one model from another. This is a profound difference between our approach and other more conventional approaches, for it means that almost all knowledge is implicit in the structure of the device that carries out the task rather than explicit in the states of units themselves. (135-136).
An ambiguity arises in Posner's overview of the volume when he writes that "cognitive sciene deals with the nature of intelligence from the perspective of computation" and references chapters 1 (written by Herbert Simon, Craig A. Kaplan), and 2 (written by Zenon Pylyshyn), he privileges serial computation by omitting any references to the connectionist views in chapter 4 (written by David E. Rumelhart). Since one might expect the differences to be mentioned, this is a puzzling omission. When speaking of the "architecture of cognition," he refers to chapter 2 written by Allen Newell, Paul S. Rosenbloom, and John E. Laird, adding that their model "has been and is currently of great influence in the field" (x). These authors simply note in passing that "Chapter 4 by Rumelhart covers nonsymbolic architectures, or more precisely the particular species under investigation by connectionist" (93). Posner does consider the connectionist approach to attention by Alan Allport important in "tieing the findings made with cognitive methods to underlying neural systems" (xii) and Allport references Rumelhart (632). This ambiguity is indeed puzzling in as much as Posner's research on attention is at the intersection of cognitive psychology and neuroscience. Aside from Rumelhart's remarks about the "profound difference" in the two models of the mind, the volume juxtaposes the two views as alternatives in a somewhat neutral manner.
This volume reflects the potential for a controversy. At this stage the competing models are most often termed serial and parallel computations. Simon and Kaplan offer both types of simulation in the discussion of CS methods contrasting STUDENT (serial) with READER (parallel) computer simulations as instances of the two methods (29ff.). The controversy between computational and connectionists models of the mind that structures Michael Lawson's 1998 Understanding CS seems yet to erupt.
In their "Symbolic Architectures for Cognition," Allen Newell, Paul S. Rosenbloom, and John E. Laird conclude by identifying areas for future research directions with respect to a computational model of the mind:
- how abilities are acquired
- living in a social community
- how biology is related to growth
- the relationship of emotion, feeling, and effect to cognition (127)
Given the competing architectures of the mind, a future direction would be to find a resolution. From another perspective the same issue surfaces—the relations between neurobiological data and computer simulations is "by no means clear or settled" ("Brain and Cognition," Foundations of CS, 343).
Steven Pinker raises the question: "Is language a distinct 'mental organ,' with principles of organizaition not shared with other cognitive systems such as perception, motor control, or reasoning (Chomsky/Fodor)? Or does lnaguage arise from general principles of intelligence applied to the problem of communicating with other humans over an auditory channel (Anderson/Minsky)? This issue is central to the development of Cognitive Linguistics which takes the latter position in the late 80s and 90s (Langacker/Lakoff).
Zenon W. Pylyshyn concludes "Computing in CS" with sme observations on "The Domain of CS":
What makes some area of study a natural scientific domain is the discovery (not the stipulation) that some relatively uniform set of principles can account for phenomenon in that domain. It is never the case that we can stipulate in advance preccisely what will fall into that natural domain. Nor can we stipulate in advance what the class of principles is that will define the domain; the evolution of the boundaries of a scientific domain is a gradual process, requiring provisional conjectures as one proceeds.
Cognitive science has been viewed as the study of the natural domain of cognition, where dognition includes prototypical phenomena of perception, problem solving, reasoning, learning, memory, and so on. (Foundations of CS, 85)
The study of communication, as we have inherited it institutionally, can be said to begin with Wilbur Scrhamm's discovery that the questions raised by Harold Lasswell's communication model—"Who says what, to whom via what channels, with what effect?" could only be answered with respect to specific situations. Working in what would become the Office of War Information, the exigencies of persuading the American public to support the war effort could only be dealt with by analyzing the situation..
It is worth noting that Harold Lasswell was conducting content analyses of Allied and Axis propaganda messages.
Weiner's work on "feedback" introduced a way of probing the audience in the situation. Schramm helped draft Roosevelt's fireside chats, giving him direct experience of what needed to be said to whom via the radio in order to produce the effect of supporting the war effort in a specific way—buying war bonds, not buying nylons from the black market..
The underlying principle in these early studies was simply that "the exigencies of the situation had to be acknowledged in the text of the message for its favorable reception." This was not "stipulated in advance," but was discovered by Schramm working in the WWII Office of Facts and Figures. (A History of Communication Study, (12-13). The natural domain was "situated exchanges of meaning" or communication. An example of this "principle" is conveyed in Grosz, Pollack, and Sidner's essay on "Discourse" in Foundations of CS:
Plan recognition in conversation is a feasible task Precisely because the speaker intends the hearer to perform it. The speaker cannot achiever her intended effect unless the hearer recognizes the speaker's plan; hence the speaker will include in the utterance what she believes to be sufficient information to make plan recognition possible for the hearer. (451-452)
The principle underlying communication study could be formulated as follows: the hearer's "interests" in the situation have to be articulated in the text/message for it to be channeled efficacious. Knowledge of the situation, the audience, the contraints imposed by the medium, and the speaker's intention inform the text/message. This is the implication of a research model that provokes the questions: "Who says what, to whom via what channels, with what effect?"
jjs
Notes:
_ . In their overview of CS, Simon and Kaplan in include a section on "Methods of Cognitive Research." One of the methods they identify is "content analysis" (27). This is a method that has been used in communication studies from its outset. ... 
← previous | next →
last revised:
September 13, 2007
Send comments to jjs.
copyright © jjs, 2007