Associative Learning Benchmarks
In this project, we list benchmark phenomena for associative learning, collated by a team of associative learning researchers in a pre-print manuscript (see contributors there). Here, phenomena are repeatedly observed data patterns in clearly specified procedures. Benchmarks are phenomena that are sufficiently robust and generalizable to merit evaluating theories in terms of whether they are consistent with them. Benchmarks are graded in terms of importance (AA, A, B, C) according to the criteria outlined below.
We put together these benchmarks to aid theorizing, and to evaluate existing theories. We believe that a common set of benchmarks is required to move the field forward. This data base should be regarded as a snapshot of the currently available evidence, seen through the lens of a particular group of researchers. For example, we have identified several instances where intuition, formed in many years of experimentation experience, contrasts with a lack of systematic investigations. Thus, we invite all scholars in this field to contribute to this project by providing additional data, references, and discussion.
To highlight factually wrong or problematic entries, please open an issue or directly create a pull request.
To discuss an entry, or to suggest reconsideration based on additional evidence, please open a discussion.
| Grade | Definition |
|---|---|
| AA | demonstrated across laboratories, highly general across species, reinforcement categories, and conditioned responses |
| A | demonstrated across laboratories, highly general within species, reinforcement categories, or conditioned responses, and demonstrated in at least one non-human species |
| B | demonstrated across laboratories, but - procedure-specific AND/OR - specific to humans |
| C | Credibly demonstrated several times, but - demonstrated within a laboratory but not across AND/OR - highly specific to procedural details AND/OR - considerable rate of null/opposing findings without clear knowledge of boundary conditions |
This github project is maintained by Dominik R. Bach University of Bonn, Centre for Artificial Intelligence and Neuroscience.