Neural decoders: saving the baby from the bathwater

In their recent article in TiCS, Vigotsky and colleagues [1] raise several critiques questioning the scientific and practical value of fMRI-based decoders. Most seriously, they argue that existing neural decoding methods inherently lack scientific validity due to the presence of nonspecific signals that confound the processes of interest. While they provide a valuable service by highlighting the insufficient attention paid to validity tests of decoding models, their statement that ‘decoding models cannot disentangle neural mechanisms from their epiphenomena’ represents an assumption rather than a fact and overlooks important ongoing efforts to identify and mitigate validity threats in decoding models.

Read Full Article (External Site)