Abstract

The goal of mapping psychological functions to brain structures has a venerable history. With the advent of neuroimaging techniques, this elusive goal regained vigor and became the main purpose of cognitive neuroscience. Unfortunately, as the field continues to develop, the ideal of finding one-to-one mappings from psychological functions to brain areas looks increasingly unrealistic. In the past few years, however, many cognitive neuroscientists have advocated for mining large sets of neuroimaging data in order to find the elusive one-to-one mapping. One recent strategy, proposed by Genon and colleagues (2018), constitutes one of the most concrete proposals for discovering the mappings from brain regions to cognitive functions by using big-data repositories of neuroimaging results. In this paper we offer several challenges for their proposal and argue that big-data approaches to finding one-to-one mappings between brain regions and cognitive functions suffer from significant difficulties of their own.