Basic mechanisms for graded persistent activity: discrete attractors, continuous attractors, and dynamic representations.

摘要: Persistent neural activity is observed in many systems, and thought to be a substrate for holding memories over time delays of few seconds. Recent work has addressed two issues. First, how can networks neurons robustly hold such an active memory? Computer systems obtain significant robustness noise by approximating analogue quantities with discrete digital representations. In similar manner, theoretical models persistent spiking have shown that the most robust stable way store short-term memory continuous parameter approximate it representation. This general idea applies very broadly mechanisms range from biochemical single cells large circuits neurons. Second, why commonly cortex strongly time-varying? observation almost ubiquitous, therefore must taken into account our understanding are held cortex.