Community Research and Development Information Service - CORDIS


Cortic_al_gorithms Report Summary

Project ID: 339490
Funded under: FP7-IDEAS-ERC
Country: Netherlands

Periodic Report Summary 2 - CORTIC_AL_GORITHMS (Cortical algorithms for perceptual grouping)

Most theories hold that early visual cortex is responsible for the local analysis of simple features while cognitive processes take place in higher areas of the parietal and frontal cortex. However, our findings now also implicate early visual cortex in tasks where subjects reason about what they see. Our results imply that the modulation of neuronal activity in early visual areas has a causal role in cognition. These influences allow the early visual cortex to act as a multiscale cognitive blackboard for read and write operations by higher visual areas, which can thereby efficiently exchange information.
We developed a number of visual cognitive tasks that can be carried out by mice, monkeys and humans. One example is a contour grouping task, where subjects see a number of curves and it is their task to mentally trace one curve (target curve) and to ignore the other curves (distractor curves). We find that when subjects mentally trace a target curve, the representation of this curve is enhanced in areas of the early visual cortex. The cortex consists of six layers and we observed a characteristic profile of top-down inputs in the superficial layers and layer 5 of the primary visual cortex, causing an increase in the firing rates in these layers, which receive feedback from higher visual areas. We also observed working memory signals in area V1. These results provide new insights in the role of early visual cortex as a cognitive blackboard that supports the implementation of mental programs.
We also examined the role of feedback connections in learning. At first sight, a neuron seems to only have access to information about its own activity as well as activity of neurons from which it receives input. Its influence to the behavior of the organism is often indirect and depends on multiple multi-synaptic pathways. Feedback signals from the response selection stage can highlight a part of the cortical network for plasticity, making the respective synapses sensitive to neuromodulatory signals that encode the “reward prediction error”, which is the difference between the amount of reward that was predicted and the amount that was obtained by the individual. The combined action of feedback connections and the reward prediction error change the synapse to promote future actions leading to more reward. The resulting learning rule is powerful and can explain how learning occurs in deep neurobiological networks.

Reported by

Follow us on: RSS Facebook Twitter YouTube Managed by the EU Publications Office Top