The binding problem

The binding problem in cognitive science refers to the lack of a known mechanism for decomposing and associating information in fixed biological neural networks.1 Several limitations of artificial neural networks are also directly related to binding.2

Here are two proposed solutions.

Analog fading memory

Synchronized spiking

A third mechanism has been proposed, involving molecular computation inside neurons. IMO this is the most interesting hypothesis so I’ll split it into a separate note.

Back to index


  1. Roskies AL. The binding problem. Neuron. 1999 Sep;24(1):7-9, 111-25. doi: 10.1016/s0896-6273(00)80817-x. PMID: 10677022. ↩︎

  2. Greff, Klaus, Sjoerd van Steenkiste, and Jürgen Schmidhuber. On the Binding Problem in Artificial Neural Networks. arXiv preprint arXiv:2012.05208 (2020). ↩︎

  3. Maass, W., Natschläger, T., & Markram, H. (2004). Fading memory and kernel properties of generic cortical microcircuit models. Journal of Physiology-Paris, 98(4-6), 315–330. doi:10.1016/j.jphysparis.2005.09.020 ↩︎

  4. Shadlen MN, Movshon JA. Synchrony unbound: a critical evaluation of the temporal binding hypothesis. Neuron. 1999 Sep;24(1):67-77, 111-25. doi:10.1016/s0896-6273(00)80822-3. PMID: 10677027. ↩︎