The Role of Computation in Psychological Explanations Essay

The Role of Computation in Psychological Explanations

     Goel looks into the capability of the Computational Theory of the Mind (CTM) to provide cognitive science with a way to comprehend and explain the representational capabilities of the human mind (1995).  He notes that that the computational concept is insufficient in totally representing the thought processes that occur in the mind because these mental representations or pictures are sometimes ambiguous and fail to satisfy the discrete, rigid and accurate conceptions required in computation (Goel, 1995). Nevertheless, there are times when the CTM concept is applicable and it is during these times that it could lend a hand in psychological explanations.

     The author starts with a discussion of folk psychology, which tends to assign mental representations to a person in the form of beliefs, desires, fears and other characteristics that are believe to be the bases for human actions (Goel, 1995). He notes that while certain disciplines, such as history and sociology, accept such explanations, scientific psychology cannot because it has to elucidate on the nature of the mental states and it must also provide an explanation for unconscious processes and seemingly irrational behavior. It is therefore attractive to the psychologist and cognitive scientist to have computational systems being representative of the workings of the human mind.

     Goel discusses two kinds of representational concepts of the mind, namely, the Intentional Theory of Mind (ITM) and its diluted version known as the Representational Theory of Mind (RTM) (1995). He describes the constraints for the ITM and the RTM concepts that he claims should be satisfied by the computational theory if it has to be accepted for explaining the workings of the mind. He also looks into the CTM properties and argues that the CTM concept is only able to comply with “epistemic” alternatives for the above-mentioned constraints, or the RTM’-constraints (Goel, 1995).

     In ITM, Goel believes that the specific processes and mechanisms should be physically achievable, it should provide a causal connection between the semantic contents and attitudes and

The Role of Computation in Psychological Explanations 3

the rationalization of human behavior, and it should be able to quantify over the semantic contents and attitudes found in  the mental states (1995). RTM is a diluted version of ITM in that the attitudes are removed from consideration. However, despite the reduction in the requirements or constraints in RTM, Goel notes that the CTM that is derived from it is unacceptable as a computational model for thought processes because computers are incapable of satisfying the mentioned requirements. He, therefore, proposes an adjusted list of requirements, which are the RTM’-constraints.

     In the RTM’-constraints, Goel maintains the need for physical realizability but makes adjustments to the two other constraints. Instead of the need to quantify the contents of the mental states he proposed that the system must allocate the contents to the computational states. Instead of the requirement to have a causal relationship between behavior and contents, there is now just a need to have a causal and non-arbitrary capability to interpret computational states as the system develops over time.

      Goel then elucidates on the characteristics or properties of the Computational Theory of the Mind or CTM (1995). The first is the requirement of syntactic disjointness of the equivalence classes of the various physical states that make up the computational states. The second property is a “causally efficacious syntax,” which means that the functioning of the computational system must be causally dependent on certain characteristics, such as size or shape (Goel, 1995 p. 54). Another required property of CTM is syntactic differentiation, which means that the computational system must be capable of distinguishing the various states. It is observed here that differentiability is restricted by the capability to design and produce sensory instrumentation for distinguishing one state from the other. Another essential CTM property is unambiguity, which is obviously required for the computational system to make the appropriate decision or interpretation or else, its output would be random in nature and would have no significance. Another obvious CTM requirement is that every state in the analysis of the computational system should have the proper causal

The Role of Computation in Psychological Explanations 3

relationship. The second-to-the-last property is semantic differentiation, which means that the system should be able to classify where each entity should belong in the compliance classes. The last requirement is the preservation of the various requirements as the system is transformed from a certain instantaneous description to another.

     Goel notes that for a computational system to comply with the RTM-constraints, it should be able to answer a set of difficult questions that include the areas of mental causation and semantic content (1995). CTM’s capability to answer these questions is tested because computational systems are known to deal with representations and no mystery is involved on how they function. This is

plausible because the mechanisms that permit computers to employ symbols in information processing might be the mechanisms that also explain the capability of the mind to process cognitive or representational data.

     Unfortunately, Goel finds that computational systems are unable to provide acceptable answers to issues about content and reference (1995). Moreover, they are incapable of handling semantic characteristics. Nevertheless, the use of computational systems to investigate the capabilities of the mind is not entirely problematic because some systems are indeed able to comply with the RTM’-constraints (Goel, 1995). He notes that CTM-compliant systems are helpful for problems that are not ambiguous or vague. But for the other problems, he posits that there is a need for non-CTM systems.

     So what then is the role of computational systems with regards to psychological explanations? In the area of CTM models, their capabilities are very limited because of their inability to represent how the mind works in cases where there is ambiguity and vagueness. It is unfortunate that much of human activity involves such problems, such as in the arts, sketching, poetry, and painting. This is not surprising in view of the fact that humans still know very little about the workings of the brain and the mind. Nevertheless, the CTM model is a step, albeit a small step, towards explaining the inner mechanisms of the mind.

The Role of Computation in Psychological Explanations 3

     What is the significance or importance of the so-called CTM properties? These are the characteristics that define the capabilities or limitations of a computational system as it attempts to e emulate the workings of the representational capacity of the mind. One example is the unambiguity property, which Goel has noted to be a severe deficiency in the CTM model (1995). For a system to work, the various physical states that it must work with have to be unambiguous or else, how would its logical processes be able to work? A computer program is based on a series of decision-making activities at each juncture, and the computational system must be able to decide on what course of action it would take. Thus, it is important that it would be able to unambiguously classify a physical state for it to be able to correctly perform its work. In psychological explanations, this CTM property also makes this model to be severely limited because of the ambiguous nature of some thought processes and resulting actions.

     Another important CTM property is the need for the system to be able express similar classes of computational states as a function of causally related properties, such as size or shape. This particular characteristic will be quite helpful in psychology as long as the problem is well-structured (Goel, 1995). In problems that involve specific features, such as shapes or sizes, the CTM model will prove be useful in the field of scientific psychology. However, once again, there are certain shapes that are difficult to classify such as those found in paintings or sketches. Here, the CTM model fails again to be helpful in psychology and the cognitive sciences.

     Another essential CTM property is syntactic disjointness or the characteristic that each computational state must be disjoint from the other states of different equivalence classes. In other words, there must be no overlap between various equivalence classes. Once again, this is important for a computational system to be able to make a decision. The applicability of this CTM characteristic is also limited in psychology. This is because in the workings of the mind there are certain situations where there is a clear distinction between the equivalence classes but there are some situations where this is vague and there are are overlaps.

The Role of Computation in Psychological Explanations 3

     Proper causal relationships in the various computational states is another essential CTM property. This is to be expected in a logical system such as a computer but this is a serious deficiency in the CTM model. It is well-known that people sometimes perform seemingly irrational actions because of emotional reasons. But perhaps, there is hope for this particular CTM characteristic by taking into account the emotions. This particular aspect is of great importance in the field of psychology and should be investigated further.

     Another significant CTM property is syntactic differentiation, which means that for every computational state in the system, it would be able to physically distinguish or differentiate each state. It should be observed that the capability of a machine or system to differentiate is restricted by the capabilities of the sensory devices. This is an important aspect in CTM and once again, it is a serious defect in the model.

     Another CTM property is semantic differentiation or the need to effectively distinguish semantically the particular compliance class of a certain entity. Goel finds that computational systems cannot provide satisfactory explanations for semantic properties nor could they give sufficient answers to issues about content and reference.

     What is the significance of these findings in the fields of cognitive science and computational systems? Goel notes that there are substantial differences between problems that have clearly defined boundaries and are not vague or ambiguous with those that are often encountered in the arts such as sketching, poetry, and painting (1995). In these so-called ill-structured problems, systems of external symbols that do not comply with the CTM-properties are needed to assist in some of the cognitive processes.

     Goel explains that there is a need to accept a psychological reality in which these symbol systems that do not comply with the CTM properties are present and that this means a need to look for mental representations of the non-CTM-type (1995). The author admits that the conclusions that can be drawn from his work are mostly in the negative area and that the current concepts of

The Role of Computation in Psychological Explanations 3

computational  systems are insufficient to completely represent all of the possibilities of the symbolic activities performed by the human mind.

     The author points out, however, that the findings do not signify the end for computationalism (Goel, 1995). What has been proven to be inadequate is the CTM model that has been detailed in the previous discussions. He expresses hope that there might be a concept of computational systems that would be able to totally encompass the characteristics of human cognition. Only time will tell when researchers would be able to come up with such a theory.

     Currently, the theories that have been presented by various scientists and researchers fail to stand up to the measure required by the workings of the human mind. Psychologists will be dismayed at the conclusions of this book although many would have expected them. At the moment, they would have to be satisfied with utilizing the CTM model as a way to emulate certain aspects of the human mind.

Reference List

Goel, Vinod. (1995). Sketches of Thought. Cambridge, Massachusetts: The MIT Press.