Brain-Score is a collection of benchmarks: combinations of data and metrics that score any model on how brain-like it is.

Data is organized in BrainIO, metrics and benchmarks are implemented in this repository, and standard models are implemented in candidate-models.

The primary method this library provides is the score_model function.

brainscore.score_model(model_identifier, benchmark_identifier, model)[source]

Score a given model on a given benchmark. The model needs to implement the BrainModel interface so that the benchmark can interact with it. The benchmark will be looked up from the benchmark_pool and evaluates the model on how brain-like it is under that benchmark’s experimental paradigm, primate measurements, comparison metric, and ceiling. This results in a quantitative Score ranging from 0 (least brain-like) to 1 (most brain-like under this benchmark).

The results of this method are cached by default (according to the identifiers), calling it twice with the same identifiers will only invoke once.

  • model_identifier – a unique identifier for this model

  • model – the model implementation following the BrainModel interface

  • benchmark_identifier – the identifier of the benchmark to test the model against


a Score of how brain-like the candidate model is under this benchmark. The score is normalized by this benchmark’s ceiling such that 1 means the model matches the data to ceiling level.