Computationalism | Is Your Brain Just a Complex Computer?

Computationalism, also known as the computational theory of mind, posits that the human mind or brain is an information processing system and that thinking is a form of computation. In this view, the brain functions similarly to a digital computer. It receives input from the senses, processes this information according to systematic rules, and produces output in the form of behavior or conscious thought. A key concept here is the Turing machine, a theoretical model of computation that can simulate any algorithmic process. Computationalism suggests the brain is a biological Turing machine, where neural states correspond to symbols and neural processes correspond to the manipulation of these symbols according to rules. This framework allows cognitive scientists to model mental processes like problem-solving, language acquisition, and perception as computational algorithms. It asserts that the mind's operations are not mysterious or non-physical but are instead grounded in systematic, rule-based symbol manipulation, making cognition a subject that can be scientifically studied and understood in principle, much like the software of a computer.

Central to computationalism is the "Language of Thought Hypothesis" (LOTH). This hypothesis proposes that thinking occurs in a mental language, often called 'Mentalese'. Just as a written language has a vocabulary (symbols) and grammar (syntax), Mentalese consists of mental representations (symbols) that are manipulated according to syntactic rules. These symbols represent concepts, and the rules governing them dictate how thoughts are formed and connected. For example, the thought 'the cat is on the mat' is composed of mental symbols for 'cat', 'mat', and the relationship 'on'. Cognitive processes are the syntactic operations that combine and transform these representations. This structure explains the productivity and systematicity of thought; we can generate a limitless number of new thoughts (productivity) and our ability to think a certain thought is intrinsically linked to our ability to think related thoughts (systematicity). According to this view, the neural firings in the brain are the physical realization of these symbolic computations.

Within the computational framework, memory is analogous to a computer's data storage. The process of encoding a new memory is like writing data to a hard drive, where sensory information is converted into a symbolic format. Retrieval is the process of accessing this stored data. Forgetting can be seen as data corruption or a failure in the retrieval algorithm. Decision-making is modeled as an algorithm that weighs variables and probabilities to arrive at a conclusion. For instance, when choosing a restaurant, the brain computes variables such as cuisine type, distance, price, and reviews, executing a program that selects the optimal choice based on predefined rules or learned preferences.

The distinction between hardware and software is a fundamental aspect of the computer analogy for the brain. The 'hardware' is the physical structure of the brain itself—the neurons, synapses, and various anatomical regions. This biological substrate is the machinery that performs the computations. The 'software' refers to the cognitive processes and mental algorithms that run on this neural hardware. These are the rules, representations, and procedures that constitute thinking, perception, and control of behavior. This separation implies that a mental state (software) is defined by its computational role, not by the specific physical material (hardware) it is implemented on, a concept known as multiple realizability.

Computationalism and connectionism are two distinct approaches to understanding the mind. Classical computationalism, as described, relies on serial processing of discrete symbols according to explicit rules, much like a traditional computer. Connectionism, on the other hand, is inspired by the brain's actual neural architecture. It proposes that mental processing occurs in parallel, distributed across a network of simple, interconnected units called nodes (analogous to neurons). In a connectionist model, knowledge is not stored in discrete symbolic representations but is encoded in the connection strengths (weights) between nodes. Learning occurs by adjusting these weights based on experience, rather than by manipulating symbols. While computationalism excels at explaining rule-based, systematic thinking, connectionism provides a more neurally plausible model for processes like pattern recognition and associative learning. Some theorists now advocate for hybrid models that integrate both symbolic manipulation and neural network processing to provide a more complete picture of cognition.