Characterizing the neural correlates of reasoning
The brain did not develop a dedicated device for reasoning. This fact bears dramatic consequences. While for perceptuo-motor functions neural activity is shaped by the input's statistical properties, and processing is carried out at high speed in hardwired spatially segregated modules, in reasoning, neural activity is driven by internal dynamics, and processing times, stages, and functional brain geometry are largely unconstrained a priori. Here, it is shown that the complex properties of spontaneous activity, which can be ignored in a short-lived event-related world, become prominent at the long time scales of certain forms of reasoning which stretch over sufficiently long periods of time. It is argued that the neural correlates of reasoning should in fact be defined in terms of non-trivial generic properties of spontaneous brain activity, and that this implies resorting to concepts, analytical tools, and ways of designing experiments that are as yet non-standard in cognitive neuroscience. The implications in terms of models of brain activity, shape of the neural correlates, methods of data analysis, observability of the phenomenon and experimental designs are discussed.
Code (0)
등록된 구현이 없습니다.
Similar Papers 제목 키워드 기반
Occipital and left temporal EEG correlates of phenomenal consciousness
In the first section, Introduction, we present our experimental design. In the second section, we characterize the grand average occipital and temporal electrical activity correlated with a contrast in access. In the thi…
EEGElectroencephalogram (EEG)Experimental DesignMeasuring AI Accountability Through Argumentation Analysis: Can Model Reasoning Withstand Scrutiny?
AI oversight methods rely on ground truth for validation, but what constitutes appropriate AI behavior is contested. This leaves evaluation of moral reasoning in LLMs and debate-based oversight implicitly avoiding realis…
On the Convergent Properties of Word Embedding Methods
Do word embeddings converge to learn similar things over different initializations? How repeatable are experiments with word embeddings? Are all word embedding techniques equally reliable? In this paper we propose evalua…
Word EmbeddingsDo Sparse Autoencoders Identify Reasoning Features in Language Models?
We study how reliably sparse autoencoders (SAEs) support claims about reasoning-related internal features in large language models. We first give a stylized analysis showing that sparsity-regularized decoding can prefere…
Characterizing Stereotypical Bias from Privacy-preserving Pre-Training
Differential Privacy (DP) can be applied to raw text by exploiting the spatial arrangement of words in an embedding space. We investigate the implications of such text privatization on Language Models (LMs) and their ten…
Language ModelingLanguage ModellingPrivacy Preserving