Introduction
Computational Theory of Mind (CTM) is widely assumed to be the primary working hypothesis in cognitive science. CTM is frequently understood as a subset of the Representational Theory of Mind, which holds that cognition is the manipulation of representation. The most widely accepted version of CTM, classical CTM, otherwise simply CTM without qualification, is related to Jerry Fodor’s Language of Thought Hypothesis (LOTH) (Jara-Ettinger, 2019).
However, several other computational profiles of the mind, most notably endosymbiont and many accounts in modern computational neuroscience, either reject LOTH or do not adhere to RTM. Furthermore, some authors explicitly distinguish between whether the brain is computational and whether it distorts representations. Even though most adherents of CTM concur that the account of learning and memory in terms of computing over illustration is the most cogent, there appears to be no inconsistency in keeping that cognition necessitates computation without adhering to representationalism (Jara-Ettinger, 2019). This paper will cover Fodor’s Computational Theory of Mind extensively. The theory is further viewed under the criticism of Searle. Finally, arguments are made against CTM in favor of Searle’s criticism.
CTM
One of CTM’s most fundamental philosophical debates is that it can demonstrate how thought and substance are causally pertinent in the physical world. It accomplishes this by stating that opinions are syntactic entities computed over. Their form renders them causation relevant in the same manner that the form renders computer source code fragments causally relevant. This basic argument can be refined in a variety of ways. Allen Newell, for example, phrased it regarding the physical symbol supposition, which holds that being a tangible symbol system is a sufficient and necessary requirement of thinking (Nayyar et al., 2022). In formalist terms, Haugeland stated that if someone starts taking care of the phrasing, the meaning will take control of itself.
Variants of CTM
The general claim that the brain is a computer can be understood in various ways based on how the basic concepts are grasped. Some theorists, in particular, contended that only cognition is computational, while emotional processes are not, even though some theorists discuss neither motor nor sensory operations in computational terms. These distinctions are minor compared to the numerous interpretations of “computation.”
Classical CTM
The ideal scenario for classical CTM can be created by demonstrating its ability to deal with abstract thinking, sound logic, and language processing. For instance –, Fodor argued that language productivity, the ability to produce an infinite number of different sentences, can only be explained by compositionality, an attribute of rich sign systems comparable to natural language (Young & Carolyn Dicey Jennings, 2022). Production systems, for example, excel at mimicking human effectiveness in mathematical and logical domains. Production restrictions, which are rules of the type “if a status X is met, do Y,” are found in production systems. In most production systems, thousands of rules are active at the same time.
However, it could be argued in his later writings that only marginal processes are computational, as opposed to core cognitive processes, which cannot be described computationally due to their holism. This negativity about classical CTM contradicts the classical approach’s successes in its significant areas. Because classical CTM is silent on the neural accomplishment of symbol systems, connectionists have criticized it as biologically implausible.
Computational Neuroscience
Computational neuroscience employs many methods, and it is challenging to find modeling techniques that apply to a broad range of task domains. However, it has been contended that, in general, brain computation is neither entirely analog nor wholly digital. This is because neurons appear to be digital on the one hand. After all, they significantly increase only if the input signal surpasses a certain threshold, causing the constant input value to become discrete.
However, their sudden increase creates continuous trends in time (Young & Carolyn Dicey Jennings, 2022). As a result, it is common to describe the operation of spiking neurons as complex systems, which implies that they are depicted in terms of consistent parameters evolving in a multidimensional space. In this case, the mathematical description shapes differential equations and networks of information-processing components (Young & Carolyn Dicey Jennings, 2022). Prototype analog/digital mechanisms are also frequently proposed to be located in various brain regions.
Connectionism
Compared to classical CTM, connectionism is typically presented as a more biologically relevant variant of computation. Even if some artificial neural networks (ANNs) are tremendously idealized for assessing neural plausibility, many researchers believe they are far more reasonable than rule-based production systems (Young & Carolyn Dicey Jennings, 2022). Connectionist systems excel at modeling motor and perceptual processes, which are significantly challenging to represent symbolically. The connection weights, in particular, are continuous values, and while these networks are typically recreated on modern systems, they are expected to incorporate analog computation. An intriguing epistemological problem emerges here: since all measurement is of finite precision, one can never be sure whether the measured value is continuous or discrete.
The discreteness could be a property of the measurement system apparatus. As a result, continuous attributes are always theoretically suggested rather than experimentally discovered because there is no way to determine whether a given value is discrete (Young & Carolyn Dicey Jennings, 2022).
Some scientific domains may have sufficient grounds to assume that estimated values should be mathematically defined as real numbers rather than digitally approximated. In cognitive science, the primary goal of computational modeling is to describe and predict psychological phenomena. Another primary goal of research in neuroscience and psychology is a therapeutic intervention (Young & Carolyn Dicey Jennings, 2022). Functionalism, particularly David Marr’s account, and mechanism are the two main conflicting theories of computational explanation. Although some in cognitive science make a case for the Deductive-Nomological account, particularly proponents of dynamics, the dynamical frameworks in question are compared with computational ones. Furthermore, the relationship between mechanical and dynamic explanation is a source of contention.
Mechanism of Computational Theory
The constitutive mechanistic explanation is the dominant form of computational interpretation in cognitive science. This type of interpretation encompasses at least three thresholds of mechanism: a constitutive (-1) level, which represents the lowest in the given analysis; an isolated (0) level, which specifies the parts of the mechanism as well as their interactions (activities or operations); and a contextual (+1) level, which places the mechanism’s function in a broader context (Nayyar et al., 2022). For instance, the context for human vision involves lighting conditions. Levels in this context are not just abstraction levels but also composition levels. They are tightly coupled but not wholly attributable to the lowest possible level.
Computational models explain how a mechanism’s computational capacity is generated by the coordinated operation of its constituent parts. To suggest that a mechanism enacts a computation means that the mechanism’s causal organization is such that the output and input information channels are intrinsically related and that this relation, alongside the particular structure of cognitive processing, is fully described. It should be noted that the connection sometimes is cyclical and can be pretty complex (Nayyar et al., 2022). The mechanistic case of computational explanation can be considered a causally constrained model of functional explanation. However, advances in mechanistic interpretation, which has become one of the most influential fields in the philosophy of science, have made it much more sensitive to modelers’ actual scientific practice.
Functionalism of Computational Theory
The modeler should ask what procedures the system functions and why it accomplishes them at the computational level. Surprisingly, the concept Marr suggested for this threshold has caused some confusion. As a result, it is usually described in semantic contexts, such as understanding or portrayal, but this can be misleading. At this stage, the data model is expected to presume that a device accomplishes a task by performing a series of processes (Nayyar et al., 2022). One must identify the task at hand and justify her informative strategy by guaranteeing that her configuration mirrors the machine’s performance and that the effectiveness is reasonable in the particular environment. The term “computation” in Marrian refers to computing tasks rather than the manipulation of specific semantics.
The extent of hardware implementation describes the physical machinery that performs the simulation; in neuroscience, this is the brain. Marr’s methodological framework is based on his computational neuroscience modeling, but it emphasizes the levels of relative autonomy, which are also layers of realization (Nayyar et al., 2022). Marr supports the conventional functionalist argument of level autonomy, which underpins antireductionism because there are multiple realizations of a specific job (Nayyar et al., 2022). Most functionalists later adopted Marr’s levels of understanding of the human brain.
Searle’s Critique of CTM
Among the most challenging issues for CTM, advocates are determining if a given physical model is an execution of a proper computation. It is noted that computer science does not provide any application theory, and the intuitive idea that identifying a link between physical and computation states may determine whether a system enacts a computation is erroneous. The following summarises Searle’s objection towards Fodor’s computational theory of mind. Physical computation is not objective; human observers ascribe computation to physical systems purely for convenience (Bora, 2020). As a result, no legitimate computational explanations of the mind exist. Nonetheless, such an objection nullifies the majority of cognitive science research.
According to Searle, being a computer system is simply assigning 0s and 1s to a physical process. For whatever program and any relatively complex object, there is an explanation of the item under which the program is realized (Bora, 2020). According to this viewpoint, even a standard wall could be a computer. In principle, both objections make the same point: given adequate freedom, it is possible to map physical states — the number of which can be adjusted logically or simply by taking more measurements — to the formal system. When individuals talk about both systems based on sets, the attribute of both sets is all that matters.
Argument
The computational theory of mind (CTM) holds that the mind is a calculation performed by the brain. To be more precise, the mind operates through the rule-based modification of symbols, just as a computer performs — computation. An individual’s states of mind are thus computational. Several prominent philosophers, including Hilary Putnam, Jerry Fodor, and, more recently, Matthias Scheutz, have held this view (Bora, 2020).
I believe that the mind’s computational model is fatally flawed. The most immediate explanation is that each mental state has meaning — they are deliberate. Intentionality implies that our opinions are always focused on something — there is constantly an item to which a thought attributes—thinking about my vacation, politics, or dog. However, computation, defined as symbol manipulation, is never profoundly about anything. A computer compares one set of electron configurations to another set of electron configurations. What those electron configurations are about is not implicit in the computation. For instance, the significance of this position as I type is not inherent in the electron patterns on my screen but in the ideas in my head.
Electron patterns or signs of any kind can represent one’s thoughts. However, it is illogical to claim that a sign is an individual’s concept (Bora, 2020). A symbol implies the thing it represents. A map of a city assumes the city; to suggest that the map is still the city is absurd. Similarly, the sequence of electrons on the screen implies the thought being expressed. Because thoughts cannot be symbols in and of themselves, thinking cannot be manipulating symbols.
Computation cannot be thought of. Searle’s position is that if thought is computation, in the context of computation, as an interrelation between states of matter, then even a blank wall contains an infinite number of quantum states of matter (Bora, 2020). Supposing that thoughts are computations, it can thus be said to enact any computation, including my own. Nonetheless, many neuroscientists currently argue that the brain is a computer. However, with a sufficiently broad description of computation, one could argue that brain states resemble computational processes in some ways. It is a poor metaphor, but much of modern neuroscience is a collection of poor metaphors about the brain and mind.
Conclusion
CTM is an effort to address mental issues by positing a syntax of thought. The move has many appeals. It seems to simplify two challenges, psychological meaning and symbolic value, to just one. The intentional idiom seems to provide a direction of vindicating psychological explanation. However, these features of progress are deceptive. The fundamental problem is that the concepts of symbols, syntax, and concepts are paradigm-driven. The conceptual occurrences are all convention- and intention-laden at their core. Any hope for a practical interpretation of these semantics of states of mind, on the contrary hand, depends on providing sufficient conditions for semantic attributes in a way that does not rely on perceptions of convention or purpose.
Whatever the computational model’s limited value in understanding the brain, CTM is a gross mischaracterization of the mind. Not computation, but the mind. However, it is the polar opposite of computation in that states of mind are always intentional, whereas computation is never intentional. Thought always has meaning, whereas computation never has its meaning. The only significance of computation as a theory in comprehending the mind is understanding that the salient features of computation and also the salient features of thought do not have points of convergence at all. Computation reveals what the mind is not and tends to inform the world of the perceived human brain functioning. The notion of computation is beneficial in the research of the mind only in that sense.
References
Bora, M. (2020). Intentionality, understanding, and symbol grounding: Searle’s Chinese room argument and the limits of computationalism. Ir.nbu.ac.in, 16. Web.
Jara-Ettinger, J. (2019). Theory of mind as inverse reinforcement learning. Current Opinion in Behavioral Sciences, 29, 105–110. Web.
Nayyar, A., Kumar, S., & Agrawal, A. (2022). Applications of computational science in artificial Intelligence. In www.igi-global.com. IGI Global. Web.
Young, B. D., & Carolyn Dicey Jennings. (2022). Mind, Cognition, and Neuroscience. Routledge. Web.