[PDF] The neuroconnectionist research programme | Semantic Scholar (2024)

[PDF] Semantic Reader

This paper presents a comprehensive review of current brain-inspired learning representations in artificial neural networks, and investigates the integration of more biologically plausible mechanisms, such as synaptic plasticity, to improve these networks’ capabilities.

A typology of model-based approaches from computational cognitive neuroscience is adopted, providing an overview of the study of mismatch responses and their alterations in schizophrenia from four complementary perspectives: connectivity models, decoding models, neural network models, and cognitive models.

Computational analyses were able to reveal PS’s representational deficits in high-level visual and semantic brain computations, highlighting the complex nature of processes contributing to face recognition in the human brain.

It is argued that the explanatory scope of neural mechanisms can be approximated by effect sizes collected across studies, not just conceptual analysis, and synthesize a mechanistic agenda that allows subfields to connect at the level of theory.

Given the large amount of recently published work in the `computational cognitive neuroscience' community, it is believed that this survey nicely organizes the plethora of work and presents it as a coherent story.

It is argued that theorists interested in developing biologically plausible models of human vision need to direct their attention to explaining psychological findings and theorists need to build models that explain the results of experiments that manipulate independent variables designed to test hypotheses.

It is argued that fMRI is well-placed to integrate the diverse subfields of systems, cognitive, computational and clinical neuroscience and can help usher in a new era of interdisciplinary coherence in neuroscience.

It is found that seRNNs, similar to primate cerebral cortices, naturally converge on solving inferences using modular small-world networks, in which functionally similar units spatially configure themselves to utilize an energetically-efficient mixed-selective code.

...

...

It is argued that the neuroscientific interpretability of ANNs can be pursued in parallel with, but independently from, the ongoing efforts in AI.

Various aspects of brain-constrained neural models may need to be constrained to make them more neurobiologically realistic and therefore better tools for understanding brain function.

NeuroGym is an open-source Python package that provides a large collection of customizable neuroscience tasks to test and compare network models, and possesses a shared interface tailored to common needs of neuroscience tasks that facilitates their design and usage.

A road map of how neuroscientists can use deep networks to model and understand biological brains is offered to offer a road map for systems neuroscience research in the age of deep learning.

It is argued that the main problem is that the authors often do not understand which human functions need to be modeled and, thus, what counts as a falsification, and DNNs can make better use by adopting an approach of comparative biology to improve the understanding of visual information processing in general.

It is demonstrated that deep learning-based models of brain circuits may be more strongly driven by particular, non-fundamental, and post-hoc implementation choices than fundamental truths about neural circuits or the loss function(s) they might optimize.

It is argued that the concept of representation provides a useful link between dynamics and computational function and asked which aspects of brain activity should be analyzed to achieve a representational understanding.

Deep neural networks represent a powerful framework for building task-performing models and will drive substantial insights in computational neuroscience.

It is found that the most powerful “transformer” models predict nearly 100% of explainable variance in neural responses to sentences and generalize across different datasets and imaging modalities (functional MRI and electrocorticography).

...

...