The neuroconnectionist research programme.

Area of Science:

• Computational neuroscience
• Philosophy of science

Background:

• Artificial neural networks (ANNs) are increasingly used to model brain and behavioral data, termed 'neuroconnectionism'.
• Current ANNs face criticism for limitations in explaining fundamental cognitive functions, prompting debate on their utility.

Purpose of the Study:

• To reframe the assessment of neuroconnectionism beyond specific ANN models.
• To present neuroconnectionism as a research program using ANNs as a computational language for brain theories.

Main Methods:

• Adopting a philosophy of science perspective, particularly Lakatos's concept of research programs.
• Analyzing neuroconnectionism's core principles, computational framework, and hypothesis-testing tools.
• Reviewing historical and current neuroconnectionist projects and their challenges.

Main Results:

• Neuroconnectionism is characterized as a progressive research program.
• The program effectively uses ANNs to formulate falsifiable theories of brain computation.
• It demonstrates a capacity for generating novel and previously inaccessible insights into brain function.

Conclusions:

• The value of neuroconnectionism lies in its potential as a research program, not solely in current ANN models.
• This approach provides a robust framework for advancing our understanding of brain computation.
• Neuroconnectionism is a dynamic and productive field for neuroscientific discovery.