2/8 There is currently no consensus on whether AI is transient or long-lasting, with studies pointing to an effect in the time scale of hs to days. These inconsistencies may obey to the method used to quantify performance, which confounds changes in retention and learning rate
— Vdellamaggiore (@vdellamaggiore) March 9, 2020
4/8 Our empirical and model-based approaches allowed us to measure the capacity for new learning independently from the influence of a previous memory. We predicted that if AI arises from an impairment in the ability to learn, adaptation in A would reduce the learning rate in B.
— Vdellamaggiore (@vdellamaggiore) March 9, 2020
6/8 Despite this strong initial bias, however, the ability to learn recovered with the passage of time, with release from interference occurring around 6 h post adaptation. pic.twitter.com/KwiWF5zqS1
— Vdellamaggiore (@vdellamaggiore) March 9, 2020
8/8. Our work shows that when adapting to conflicting perturbations, impairments in performance are driven by two distinct mechanisms: a long-lasting bias that hinders initial performance, and a short-lasting AI that originates from a reduction in error-sensitivity.
— Vdellamaggiore (@vdellamaggiore) March 9, 2020