Skip to main content
Social Sci LibreTexts

7.10: From Classical Marks to Hybrid Theories

  • Page ID
    41277
  • Vera and Simon’s (1993) analysis of situated action theories defines one extreme pole of a continuum for relating different approaches in cognitive science. At this end of the continuum, all theories in cognitive science—including situated action theories and connectionist theories—are classical or symbolic in nature. “It follows that there is no need, contrary to what followers of SA seem sometimes to claim, for cognitive psychology to adopt a whole new language and research agenda, breaking completely from traditional (symbolic) cognitive theories” (p. 46).

    The position defined by Vera and Simon’s (1993) analysis unites classical, connectionist, and cognitive science under a classical banner. However, it does so because key terms, such as symbolic, are defined so vaguely that their value becomes questionable. Critics of their perspective have argued that anything can be viewed as symbolic given Vera and Simon’s liberal definition of what symbols are (Agre, 1993; Clancey, 1993).

    The opposite pole of the continuum for relating different approaches in cognitive science is defined by theories that propose sharp differences between different schools of thought, and which argue in favor of adopting one while abandoning others (Chemero, 2009; Fodor & Pylyshyn, 1988; Smolensky, 1988; Winograd & Flores, 1987b).

    One problem with this end of the continuum, an issue that is the central theme of the current chapter, is that it is very difficult to define marks of the classical, features that uniquely distinguish classical cognitive science from competing approaches. Our examination of the modern computing devices that inspired classical cognitive science revealed that many of these machines lacked some of the properties that are often considered marks of the classical. That is, it is not clear that properties such as central control, serial processing, local and internal representations, explicit rules, and the cognitive vocabulary are characteristics that distinguish classical theories from other kinds of models.

    The failure to find clear marks of the classical may suggest that a more profitable perspective rests somewhere along the middle of the continuum for relating different approaches to cognitive science, for a couple of reasons. For one, the extent to which a particular theory is classical (or connectionist, or embodied) may be a matter of degrees. That is, any theory in cognitive science may adopt features such as local vs. distributed representations, internal vs. external memories, serial vs. parallel processes, and so on, to varying degrees. Second, differences between approaches may be important in the middle of the continuum, but may not be so extreme or distinctive that alternative perspectives cannot be co-operatively coordinated to account for cognitive phenomena.

    To say this differently, rather than seeking marks of the classical, perhaps we should find arcs that provide links between different theoretical perspectives. One phenomenon might not nicely lend itself to an explanation from one school of thought, but be more easily accounted for by applying more than one school of thought at the same time. This is because the differing emphases of the simultaneously applied models may be able to capture different kinds of regularities. Cognitive science might be unified to the extent that it permits different theoretical approaches to be combined in hybrid models.

    A hybrid model is one in which two or more approaches are applied simultaneously to provide a complete account of a whole phenomenon. The approaches might be unable to each capture the entirety of the phenomenon, but—in a fashion analogous to coarse coding—provide a complete theory when the different aspects that they capture are combined. One example of such a theory is provided in David McNeill’s (2005) Gesture And Thought.

    McNeill (2005) noted that the focus of modern linguistic traditions on competence instead of performance (Chomsky, 1965) emphasizes the study of static linguistic structures. That is, such traditions treat language as a thing, not as a process. In contrast to this approach, other researchers have emphasized the dynamic nature of language (Vygotsky, 1986), treating it as a process, not as a thing. One example of a dynamic aspect of language of particular interest to McNeill (2005) is gesture, which in McNeill’s view is a form of imagery. Gestures that accompany language are dynamic because they are extended through time with identifiable beginnings, middles, and ends. McNeill’s proposal was that a complete account of language requires the simultaneous consideration of its static and dynamic elements.

    McNeill (2005) argued that the static and dynamic elements of language are linked by a dialectic. A dialectic involves some form of opposition or conflict that is resolved through change; it is this necessary change that makes dialectic dynamic. The dialectic of language results because speech and gesture provide very different formats for encoding meaning. For instance,

    in speech, ideas are separated and arranged sequentially; in gesture, they are instantaneous in the sense that the meaning of the gesture is not parceled out over time (even though the gesture may take time to occur, its full meaning is immediately present). (McNeill, 2005, p. 93)

    As well, speech involves analytic meaning (i.e., based on parts), pre-specified pairings between form and meaning, and the use of forms defined by conventions. In contrast, gestures involve global meaning, imagery, and idiosyncratic forms that are created on the fly.

    McNeill (2005) noted that the dialectic of language arises because there is a great deal of evidence suggesting that speech and gesture are synchronous. That is, gestures do not occur during pauses in speech to fill in meanings that are difficult to utter; both occur at the same time. As a result, two very different kinds of meaning are presented simultaneously. “Speech puts different semiotic modes together at the same moment of the speaker’s cognitive experience. This is the key to the dialectic” (p. 94).

    According to McNeill (2005), the initial co-occurrence of speech and gesture produces a growth point, which is an unstable condition defined by the dialectic. This growth point is unpacked in an attempt to resolve the conflict between dynamic and static aspects of meaning. This unpacking is a move from the unstable to the stable. This is accomplished by creating a static, grammatical structure. “Change seeks repose. A grammatically complete sentence (or its approximation) is a state of repose par excellence, a natural stopping point, intrinsically static and reachable from instability” (p. 95). Importantly, the particular grammatical structure that is arrived at when stability is achieved depends upon what dynamic or gestural information was present during speech.

    McNeill’s (2005) theory is intriguing because it exploits two different kinds of theories simultaneously: a classical theory of linguistic competence and an embodied theory of gestured meaning. Both the static/classical and dynamic/ embodied parts of McNeill’s theory are involved with conveying meaning. They occur at the same time and are therefore co-expressive, but they are not redundant: “gesture and speech express the same underlying idea unit but express it in their own ways—their own aspects of it, and when they express overlapping aspects they do so in distinctive ways” (p. 33). By exploiting two very different approaches in cognitive science, McNeill is clearly providing a hybrid model.

    One hybrid model different in nature from McNeill’s (2005) is one in which multiple theoretical approaches are applied in succession. For example, theories of perception often involve different stages of processing (e.g., visual detection, visual cognition, object recognition [Treisman, 1988]). Perhaps one stage of such processing is best described by one kind of theory (e.g., a connectionist theory of visual detection) while a later stage is best described by a different kind of theory (e.g., a symbolic model of object recognition). One such theory of seeing and visualizing favoured by Pylyshyn (2003c, 2007) is discussed in detail as an example of a hybrid cognitive science in Chapter 8.

    • Was this article helpful?