A Brain-inspired Computational Model for Human-like Concept Learning (2401.06471v1)

Abstract: Concept learning is a fundamental aspect of human cognition and plays a critical role in mental processes such as categorization, reasoning, memory, and decision-making. Researchers across various disciplines have shown consistent interest in the process of concept acquisition in individuals. To elucidate the mechanisms involved in human concept learning, this study examines the findings from computational neuroscience and cognitive psychology. These findings indicate that the brain's representation of concepts relies on two essential components: multisensory representation and text-derived representation. These two types of representations are coordinated by a semantic control system, ultimately leading to the acquisition of concepts. Drawing inspiration from this mechanism, the study develops a human-like computational model for concept learning based on spiking neural networks. By effectively addressing the challenges posed by diverse sources and imbalanced dimensionality of the two forms of concept representations, the study successfully attains human-like concept representations. Tests involving similar concepts demonstrate that our model, which mimics the way humans learn concepts, yields representations that closely align with human cognition.

• The paper presents a brain-inspired model that uses spiking neural networks to mimic human concept learning by integrating multisensory and textual representations.
• The model is divided into three specialized modules—multisensory, text-derived, and semantic cooperation—to replicate distinct cognitive functions.
• Evaluations with similar concepts tests confirm the model's alignment with human cognition and its potential to advance brain-inspired AI research.

Introduction

Concept learning is a cornerstone of human cognition, fundamental to activities such as categorizing, reasoning, and decision making. In understanding concept learning, multisensory representation and text-derived representation stand out as two pillars within the brain's knowledge structure. The harmonious interplay between these pillars is orchestrated by a semantic control system within the brain, crafting a nuanced and adaptive learning process.

Mimicking Human Cognition

Researchers have created a brain-inspired computational model that leverages spiking neural networks to replicate this dance of concept learning. Challenging issues like source diversity and imbalanced dimensionality between sensory and textual data are elegantly navigated in the model, achieving a closer alignment with the way humans process and understand concepts.

Computational Model Structure

Integrating insights from computational neuroscience and cognitive psychology, the model is neatly divided into three segments: a multisensory information processing module, a text-derived information processing module, and a semantic cooperation module. Each of these mimics a specific cognitive function—corresponding to the brain's multimodal experiential system, linguistic system, and semantic control system, respectively—to construct human-like concept representations.

Model Evaluation and Future Prospects

Evaluations through 'similar concepts tests' showcase that the model generates representations resonant with human cognition. The nuanced interplay between sensory and text-based representations echoes the complexity of human thought, exemplifying the model’s sophisticated comprehension of concepts. While this research introduces a prototype aligning closely with the human thought process, future explorations in cognitive mechanisms and dataset mapping, amongst others, hold the promise of furthering brain-inspired artificial intelligence's frontier.