News & Events

Learning and attention are mechanistically distinct


Learning and attention act on different timescales, but share a common purpose: both processes improve our behavioural ability to detect, process and respond to relevant features of our environment. Previous work has shown that both learning and attention occur within the same neural circuits, and both boost the ability of neurons to signal specific stimulus features, but do they rely on the same or similar mechanisms?

In a new paper, now out in Neuron, Adil Khan, along with researchers from UCL, Cambridge and Imperial College London, addressed this question in their novel series of experiments yielding both surprising new insights and methodological advances.

Using 2-photon microscopy, Poort, Khan et al. imaged primary visual cortex (V1) in mice over a week as they learned a visual discrimination task and then, as they performed an attention-switching task over the course of a day. As mice learnt the task, and as they switched attention, individual V1 neurons showed increased selectivity, thus becoming better at signalling which stimulus was presented to the mouse. Surprisingly, even though the same neurons were involved in both processes, selectivity changes following the learning and attention tasks were not correlated. Exploring these findings, the authors found that during learning, selectivity increases were a result of suppression of responses whereas when paying attention, the increased selectivity was driven largely by selective enhancement of responses to specific stimuli.

Further exploration showed that learning and attention differentially affected interactions between excitatory cells and parvalbumin, somatostatin and VIP expressing inhibitory cells. Computational modelling of these findings, revealed that learning led to a reorganisation of local connectivity, with excitatory and PV inhibitory cells forming highly connected subnetworks. On the other hand, cell class-specific top-down inputs best explained improved attentional modulation. These findings show how different mechanisms are employed by the brain in V1 over shorter and longer time scales to improve visual performance.

Dr Adil Khan commented:

Our findings highlight that the cortex is remarkably versatile. The same individual cells can not only participate in different cognitive processes such as learning and attention, but can do so using distinct mechanisms. We can now begin to ask further mechanistic questions, such as how local connectivity changes, top-down inputs or neuromodulators guide these processes.

This important paper is a significant methodological advancement: using 2-photon imaging in behaving mice to visualise the same neurons across different behavioural tasks gives rise to a multitude of possibilities for understanding behaviour and cognition on a new level. Poort, Khan et al. focus on learning and attention in V1 and their findings prompt further work to understand behaviour and cognition at this level in other brain regions.