experiment performed by medium of computation, in the case of biology/experimental sciences it is ones in which some process can be modeled and thus simulated within the computer.
more recently has seen a redefinition for a new class of “digital twin” deep learning models, which can now serve an interesting use
one can imbue a sufficiently complex model with a Representation of some relationship: say that of a macaque Receptive Field/visual input to some source of brain activity information like a sharp Electrodes recording or fMRI activation data.
we make some assumptions that given these models are able to reliably predict brain activity given the associated stimuli, that they contain a substrate of the underlying representation that we ourselves have in the path from receptive field to information processing.
this isnt perfect, but that is part of the interesting experiment, as we get to measure divergences between model and human representations of media. (Prince et al., 2026) does this and I tried my hand at recreating it under Every Encoder Loves Its Own Images
“This paradigm promises to combine the interpretability of controlled experiments with the generalizability and broad scope of natural stimulus experiments. We show four examples of simulating language neuroscience experiments in silico and then discuss both the advantages and caveats of this approach.”
Bibliography
Prince, J. S., Wang, B., Fel, T., Jagadeesh, A. V., Vaziri, P. A., Alvarez, G. A., Livingstone, M. S., & Konkle, T. (2026). Parametric Neural Control Differentiates Top Neural Network Models of Primate Visual Cortex (p. 2026.08.16.745063). bioRxiv. https://doi.org/10.64898/2026.08.16.745063