Summary
Gilad et. al. measure neural activity from large parts of the neocortex using wide-field imaging while mice are performing a short-term memory task (i.e. whisker-based go/no-go texture discrimination task with delayed response). They find that distinct neocortical areas (frontal or posterior) maintain short-term memory depending on behavioral strategy (active or passive).
Specifically, their hypothesis about the cortical dynamics during sensation and short-term memory is the following: During sensation, a posterior network (S1BC, S2, RL) displays enhanced activity, enabling the discrimination between the two textures. At the beginning of the delay period, information flows to either frontal or posterior areas for short-term maintenance , depending on mouse’s strategy: from RL to M2 during an active strategy (motor short-term memory) or from RL to P during a passive strategy (sensory short-term memory). Both cortical states eventually converge to trigger a similar behavioural response (licking for reward).
Concerns
- What is their “movement” criterion? Usually, quiet animal = freezing, how they quantified body movements? In Fig. 1D they on mention arbitrary units.
- During sensation, they focus in S1BC, S2, RL that are activated for both strategies, but for the active strategy many more areas are activated including M2. Also, Fig. 2C for S1BC should be the same as Fig 3C for S1BC sensation period, but they are not.
- Fig. 3C,D in the delay maps of active there is always some activation in poster areas but in 3E it appears as the activation in P is almost 0.
- How can you know that the activation of M2 is indeed task dependent short-term memory related and not just motion related? Independent of the task itself and dependent just on the movements that were performed during the task? Then the sensory-related information should anyway be stored in P for active mice as well, it is just covered by the abundant motion-related activation.
- They suggest that area P processes information about object identity, receiving tactile information via RL and being part of the “what” pathway. However, the evidence so far is that P is involved in the visual processing and has nothing to do with object identity. If the experiments were done in light, it can well be that P was active due to visual information and not task dependent. They don’t mention if the experiments were performed in dark or light.