Figure 2 shows on the left side a sequence of 3 possible actions:
A, pull a lever, B, flick a switch and different possibilities for C. On top there is C1, pull a handle. Possible schematic activity patterns are shown there, too, together with potential predictions that could arise from them. For example, activity α will for many neurons be directly coupled to action A (denoted by αA), but – from prior experiments . – we expect also to find neurons where activity α under the condition that later-on action C is planned
(denoted by αγ) should be predictive for action C (denoted by αγC). More such conjunctions are shown in the
figure. The goal of Neurocientific Aspects is to record and characterize neural responses according to these possibilities.
The example shows that there are multiple combinations possible even for such a simple 3-stage action sequence.
In addition, we need to address the fact the neuronal responses will vary also between trials. To this end several
methods (see Methods section below) will be used for data analysis and interpretation.
Thus, the goal of this part of the work is:
1. To provide novel SmartCage environment for real-time behavioral
control and wireless neural recordings in unrestrained animals conduction complex action sequences.
2. To provide robust multi-channel recordings from PRR, PMd, and SMA for the above described experiment.
3. To provide solid data analyzes for characterizing different neuron types with respect to their “predictive properties”. From prior work we expect to find a large fraction of motor planning neurons; some will be selective for the immediate next movement only, irrespective of its embedding in an action sequence.