VOLUME LXXIII | JAN 2023

Stimulus models test hypotheses in brains and DNNs

In a recent response to Pitkow’s arguments, we defend our opinion on the equivalence between the brain and its DNN models. We address the concept of a ‘suitably rich’ stimulus ensemble, which Pitkow suggests as an alternative to our 2nd degree of algorithmic equivalence. Our goal is to test hypotheses and explore the relationship between the brain and artificial neural networks. Think of it like comparing two recipes for chocolate chip cookies – one made by a master chef (the brain) and the other by a sophisticated machine (DNN). By examining their similarities and differences, we gain insights into how the brain processes information and develop better DNN models. We encourage you to dive into the article to learn more about this fascinating field!

Marc Harper

Marc is a Canadian exercise physiologist in St. John’s, linking fitness to sharper thinking. He writes about real-world ways to stay strong and sharp, inspired by Newfoundland’s tough, no-frills lifestyle.

Pitkow [1] raises several points about the arguments made in our recent Opinion in Trends in Cognitive Sciences titled ‘Degrees of equivalence between the brain and its DNN models’ [2]. Here, we clarify our position before delving into the elusive argument of the ‘suitably rich’ stimulus ensemble that Pitkow suggests as an alternative to our 2nd degree of algorithmic equivalence (see also the section ‘Is the 2nd degree really necessary?’ in our article [2]).
Read Full Article (External Site)