This framework is an ongoing project.
Check the Github-Repository for framework update logs.
CCOBRA's philosophy is based around the fact that models always attempt to solve specific modeling tasks, either explicitly or implicitly. Contrary to big parts of the current state of the art in cognitive modeling, CCOBRA focuses on making the processes underlying response generation explicit. To this end, CCOBRA mandates models to revolve around a single method which asks for an explicit prediction in response to a presented task.
The sheer simplicity of CCOBRA sets it apart from most contemporary modeling approaches. Models are not required to adhere to rigorous Bayesian fundamentals nor computational logic calculi nor statistical effects extracted from years of psychological research. Instead, it encourages tackling core problems of cognitive science by imposing little to no constraints with respect to the computational foundation of model instances. The sole requirement is the ability to make predictions in response to tasks, an arguably trivial prerequisite.
CCOBRA heralds the dawn of a new paradigm of cognitive modeling — a perspective that is focused on prediction-based performance. Coupling the goal of achieving high predictive performance with rich possibilities to infer insight into the computational principles underlying model implementations, CCOBRA offers a modern toolset to aid computer scientists and cognitive scientists alike in their respective goals.
This website allows to upload models for the core tasks contained in the CCOBRA ecosystem. The top-level menu links to the domain-specific evaluation sites which allow for the upload of a user-specified model. Following the upload, the model is evaluated and its results presented.
Alternatively, all scripts related to the model definition and evaluation can be downloaded for local use: