Proceedings of the Annual Meeting of the Cognitive Science Society

Feedback alignment has been proposed as a biologically plausible alternative to error backpropagation in multi-layer per-ceptrons. However, feedback alignment currently has not been demonstrated to scale beyond relatively shallow networktopologies, or to solve cognitively interesting tasks such as high-resolution image classification. In this paper, we providean overview of feedback alignment and review suggested mappings of feedback alignment onto biological neural net-works. We then discuss a novel geometric interpretation of the feedback alignment algorithm that can be used to analyzeits limitations. Finally, we discuss a series of experiments in which we compare the performance of backpropagationand feedback alignment. We hope that these insights can be used to systematically improve feedback alignment underbiological constraints, which may allow us to build better models of learning in cognitive systems.