Are there Representations in Biological Brains?
- 1. Hui Xin, Ng Writing Sample One ofFour Essays Submitted for COGS 215: Knowledge and Cognition Response to the Claim: There are no representations in biological brains. The goal of this essay is twofold: (i) to describe how different theorists have conceptualized the notion of representation and (ii) to evaluate the utility of the concept of representation, despite its inability to fully capture the complete phenomenological experience of existence. The concept of representations is useful because it can illustrate the possible mechanisms that give rise to “meaning”. However, we must be cautious in our approach to the notion of “meaning”. Is “information” packaged in a combinatorial manner meaningful? A living thing is constantly “processing information” until the end of its life, but should we consider every single action performed during its lifetime as “meaningful”? When we study “meaning”, language is often the first topic that enters into the discussion since human beings utilize language to make sense of the world and to communicate to one another. However, the literature on embodiment theory supports the argument that meaning is not always propositional and rule-based in nature as researchers have traditionally conceived. Rather than studying meaning among adults, embodiment theory argues that the study of meaning should be anchored in the study of infants who are in the process of acquiring and understanding meaning in the world. Infant development studies are crucial in shedding light on language acquisition, specifically on the question of meaning acquisition, which is a predecessor to “full blown” use of language. As Johnson (2008) has described in the Meaning of the Body, infants have to learn the perceptual and conceptual experiences that adults take for granted – concepts such as object permanence, inertial motion and causation. Curiously enough, Johnson mentions that “conceptual capacities,” possessed by infants must be activated through experience. These capacities depend on genetic factors and developmental history of the individual infant. Before meaning can arise, proto-meanings, or immanent meanings are implicitly gleaned from interaction with the
- 2. Hui Xin, Ng Writing Sample environment. It is only from these proto-linguistic meanings that one experiences through “feeling” the patterns of the organizing processing that one can derive “meaning”. Johnson even describes adults as “big babies”; while certainly an exaggeration, the implicit claim is that the learning process in adults and infants is similar: adults do not have complete models or representations of the world in their minds. When learning a new skill or adapting to a new context, adults often attempt to make sense of things in a similar manner as infants do. In a novel situation, an adult learner must learn to understand the contextual cues and existing patterns from the environment. In the same vein of thought, Dreyfus (2004) uses the term “intentional arc” to describe the connection between the agent and the world. This is similar to Johnson’s (2008) argument that meaning arises from the interaction between the individual and its environment. The “intentional arc” that Dreyfus refers to appears to be borrowed from philosophy, except in this context the term does not seem to carry the same connotation as in philosophy; in philosophy, intentionality implies that the agent is directing a propositional attitude towards a particular object or state of affairs. Dreyfus argues that the agent responds to its current perceptions of its environment by recalling its skills it possesses that are stored as “dispositions”. Dreyfus (2004) does not provide an explanation of “dispositions” in the article, but his introduction of the concept of “maximum grip” and description of skill acquisition can help shed light on this important idea. A skilled agent moves towards a state or position that maximizes its sense of optimal gestalt. The maximum grip is the body’s tendency to respond to the beckoning of its perceptions to adjust to the situation it is in. Skills are therefore “dispositional” in nature – they are characterized as inclinations or tendencies to move towards something or to bring about an optimal state of affairs. However, it is not clear if a particular satisfactory state of affairs is ever
- 3. Hui Xin, Ng Writing Sample reached, although the agent is able to achieve contentment through “absorbed coping”; a process that allows the agent to continuously move towards an equilibrium to reduce the tension it experiences at a given moment. The tension is continuously absorbed by the “attractor landscape” – a term that is rather ill-defined by Dreyfus— which draws the agent’s nervous system to relax into a particular “attractor”. The agent never arrives at this so-called equilibrium because it is constantly moving towards a new “goal”. Amazingly, this idea, as vague as it seems, reconciles two diverging notions of “the good life” in philosophy; the Platonic vision that argues that the people act in order arrive at fulfillment, and the Aristotelian idea that the good life must necessarily entail ongoing coping and engagement with challenges. Before we launch into a discussion of how thought and meaning – and more broadly, intelligence – is embodied, we must solve the question of what an “attractor landscape” is. In the following paragraph I will explore how Johnson (2008) and Sheets-Johnstone’s (2010) arguments might shed light on the concept of “attractor landscape”. One of Johnson’s (2008) central arguments is that the meaning of causation is grounded in bodily experience. Without the basic movement of pushing and pulling, infants cannot possibly grasp the idea of causation; Johnson argues that they do feel patterns of the organizing process, before they are able to experience objects and their generic properties. This process gives rise to an emergent sense of self, which gives the infant’s experiences – both inner states and actions – form and meaning. What Johnson describes as vitality affects – the dynamic effects of experiences – is equivalent to what Sheets-Johnstone (2010) describes as the capacity for experiencing and understanding tactile kinesthetic spatiality. According to Sheets-Johnstone, the infant is able to distinguish the concepts “near” and “far” through the flow of its movement, the strains and tensions in pursuit of its goal through the experience of effort, fatigue and ease in coordinating its movement. Vitality affects, such as “gently”, “jaggedly”, and “explosively”, are closely related to
- 4. Hui Xin, Ng Writing Sample what Sheets-Johnstone names as “kinetic dynamics”. Before an infant is able to conceptualize itself as an object in space, it experiences the “kinetic dynamics” of its own movements. The infant experiences its body as an anchor for orienting itself in the world, a zero point from which it able to judge the nearness or farness between the things around it and itself. We speak of “nearness” or “farness” instead of measurable distance because the experiential space for tactile kinesthetic spatiality has nothing to do with measurable distances; rather, they are experiential dimensions of movements. Sheets-Johnstone (2010) and Johnson’s (2008) arguments are important to the aforementioned discussion of Dreyfus’ idea of an “attractor landscape” because they elaborate on Dreyfus’ description of tension resolution in the organism’s body. Using the example of sports, Dreyfus describes how the player’s comportment is a result of his/her perception. The player adjusts his/her body’s position and movement in order to reduce the deviation from an overall satisfactory gestalt. In fact, to speak of the words “in order” can be misleading since it implies that the player has a specific goal represented in his/her mind and therefore acts according to certain rules, however dynamic they are, to bring brought the action intended. The key point here is that the player often does not even “know” what the goal is, he/she only senses it as he/she is getting closer or further away from the optimum organizational comportment. There are no representations – only a constant feedback loop exists between the agent’s actions and perception of its surroundings. Given that words are metaphoric from embodied experience of movement, a lot of qualities are difficult to capture in language because qualities are felt experiences. To distill such experiences into discrete concepts is to attempt the impossible. Consider the occasion of describing one’s illness to the doctor. The doctor often asks questions to prompt the patient to specify sources
- 5. Hui Xin, Ng Writing Sample of discomfort by describing their felt experiences. Subsequently, doctors “translate” those descriptions to diagnosable symptoms. for example, “queasy”, “bloated”, “numbness”, “congestion” and prescribe medications. However, the nuances of these felt qualities might be “lost in translation”; for examples, the difference between the sensation of “tightness”, “numbness”, and “cramping” might be subtle depending how the patient defines those sensations personally. Despite the dichotomy between the discrete nature of representations and the seemingly continuous quality of our phenomenological experience, the notion of representations still dominate in many theories and models in cognitive science. Even Dreyfus (2004) describes patterns generated by the cortex as “representations” albeit in an unconventional sense; these representations stand for unique meanings of stimuli for individuals rather than the stimuli itself. In such a neural network, there are no rules or symbols, instead, there is a trained intermediate layer of nodes that forwards information to an output layer that produces output in various combinations. Subsequently, the net makes an association between new input and the output produced. If there is a wrong association, the weights on the connections between layers are changed through back propagation. Therefore, there is no literal representation per se; these representations are simply spatial patterns that are activated in the brain. Dreyfus’ account of representations enables the network to “learn” to generalize; given the task of categorizing objects, the network associates further inputs of similar type with the outputs it produces. This abstract network is instantiated in a corporeal agent. The agent, learning from its past experience, is able to refine his/her discriminations and therefore perceive its environment from an informed perspective. Nonetheless, Dreyfus’ account lacks explanatory force in drawing the connection between the neural network’s functional mechanism and the brain’s implementations. Fortunately,
- 6. Hui Xin, Ng Writing Sample Barsalou’s (2003) proposal of modal-specific simulations of concepts might be able to fill the gaps in Dreyfus’ theory. In what Barsalou names as a “simulator”, concepts emerge from the interaction between the brain and the world. More specifically, particular spatial patterns of neural activations emerge when the agent encounters a particular stimulus. These neural patterns are generated from complex interactions between modality-specific circuits from sensation, action and emotions. They emerge from the constant feedback loop from action and perception as Dreyfus describes. Concepts do not operate through an amodal system; they are emergent depending on the particular situation an agent is exposed to. In other words, concepts are situated. The “simulator” of concepts does not deliver a universal, globalized description of an object, rather, it produces a specific simulation in response to each situation, which necessarily includes the agent’s own unique mental and motor state at a given point of space and time. We are again reminded of Dreyfus’ argument that neural patterns in the brain are not representations of stimuli, but are more nuanced representations of meaning of those stimuli to the individual. Dreyfus (2004) and Barsalou’s (2003) accounts set the stage for Reed’s (1996) ideas that constitute the theories in ecological psychology. To Reed, these topographic patterns in the brain are meaningless because they do not help with increasing our understanding of how organisms “understand” the world. Even if we count those patterns as “representations” of something significant, we will still need to identify the particular aspects of representations that are relevant to the organism’s goal and task at hand. The daily experiences and activities of the organism are influential in shaping and maintaining those topographic patterns in the brain. Implicitly, the organism needs to already understand those representations in order to make use of them. With the examples of infant development from Johnson (2008) and Sheets-Johnstone and model descriptions from Barsalou (2003) and Dreyfus (2004) in mind, we may conclude that the
- 7. Hui Xin, Ng Writing Sample notion of representations seems to have been misunderstood and mischaracterized in current psychological science. As Reed (1996) succinctly summarizes in the early chapters of his Encountering the World, the source of confusion in psychology is the notion that organisms have representations of the world through collection and interpretation of stimuli and therefore have a model of the world built into them. This leads to the further problem of having to locate a “reader” inside the brain that coordinates all the information in these topographical maps. Nonetheless, representations are still important given that the promising models and theories from Barsalou and Dreyfus provide us with a new interpretation of the idea of “representation”. Discarding the idea of representations from working hypotheses in cognitive science will be a devastating blow to progress in cognitive science, even if representations cannot fully capture phenomenological experiences. We can be safely assured of the utility of the concept of “representation” in cognitive science, although we must be aware of the caveats involved; when we speak of “representations”, we must be careful about delineating what constitutes them and what their functions are.
- 8. Hui Xin, Ng Writing Sample References Barsalou, Lawrence W., et al. "Grounding conceptual knowledge in modality-specific systems." Trends in cognitive sciences 7.2 (2003): 84-91. Dreyfus, Hubert L., and S. E. Dreyfus. "A phenomenology of skill acquisition as the basis for a Merleau-Pontian non-representationalist cognitive science." Retrieved February 13 (2004): Johnson, Mark. The meaning of the body: Aesthetics of human understanding. University of Chicago Press, 2008. Reed, Edward S. Encountering the world: Toward an ecological psychology. Oxford University Press, 1996. Sheets-Johnstone, Maxine. "Thinking in movement. Further analyses and validations." Enaction: Toward a new paradigm for cognitive science (2010): 165-181.