Fig. 1. Overview of experiment and methodology.

a Experimental paradigm. In each trial, participants were presented with a picture of an object and asked to identify it; randomly occurring catch trials ensured compliance. b Overview of the brain-to-semantics mapping method. Linear regression models were (1) trained on a set of brain response-semantic vector pairs and (2) tested on the brain response to one left-out concept, (3) yielding a predicted vector. (4) The distance between the predicted vector and the target vector was calculated. This procedure was repeated for all concepts. The same procedure was performed for each instance of the cumulative window. The dashed line corresponds to the point after which prediction-target distance no longer decreased (plateau point). Semantic vectors were created using the word2vec algorithm on a large Finnish text corpus.