Beyond the "Conceptual Nervous System": Can computational cognitive neuroscience transform learning theory?

Area of Science:

• Cognitive Science
• Computational Neuroscience
• Learning Theory

Background:

• Traditional learning theory often relies on hypothetical representational nodes and connectionist approaches.
• Skinner's critique of the "Conceptual Nervous System" highlighted issues with purely hypothetical neural structures.
• Theory underdetermination, where distinct cognitive processes yield identical behavioral predictions, is a persistent problem.

Purpose of the Study:

• To demonstrate the prevalence of theory underdetermination in learning theory and cognitive modeling.
• To introduce Computational Cognitive Neuroscience (CCN) as a solution to resolve theoretical ambiguities.
• To highlight the importance of integrating neurobiological constraints into computational models.

Main Methods:

• Analysis of existing learning theory literature to identify instances of theory underdetermination.
• Examination of computational models in cognitive science and their predictive equivalence.
• Proposal of CCN models that incorporate specific neurobiological constraints.

Main Results:

• Theory underdetermination is shown to be common, hindering the resolution of key theoretical questions in learning.
• Contrasting models often produce the same behavioral predictions, obscuring fundamental differences in proposed mechanisms.
• CCN models offer a path beyond the "Conceptual Nervous System" by grounding models in neural architecture.

Conclusions:

• Computational cognitive neuroscience provides a robust framework for overcoming theory underdetermination.
• Integrating neurobiological data allows for a more precise and constrained approach to cognitive modeling.
• CCN enables a true synthesis of behavioral and neural analyses, advancing our understanding of learning.