From Size Perception to Counting - A Computational Point of View
In this study
we examine whether the counting system can develop from the size perception system through an evolutionary process, or that counting and size perception abilities are achieved by two separate systems developed in different epochs of time. We will try to computationally show whether individuals who excel in size perception have an advantage in learning to count. This will be done by using a branch of artificial intelligence termed evolutionary computation or evolutionary algorithms (EAs). It involves mechanisms inspired by biological evolution like reproduction, mutation, recombination, and selection.
Specifically, we will run a genetic algorithm (GA) in order to develop Artificial Neural Networks (ANNs) through an evolutionary process. The performance will be measured by the population sizfixing spelling errorse X number of generation it took to succeed in the given task(s).
We plan to first evolve ANNs that can learn to perceive size successfully. Then, we will examine whether those ANNs have the advantage over new learners when acquiring knowledge of counting.
Lab member: Gali Katz
