Cisek & Kalaska (2010) Neural mechanisms for interacting with the world

Cisek, P., & Kalaska, J. F. (2010). Neural mechanisms for interacting with a world full of action choices. Annual Review of Neuroscience, 33, 269-298. 

Much decision-making work in cognitive neuroscience has held an information processing perspective. In this view, there are three serial processes:

1. Perception. Take information from the outside world to build knowledge about it.
2. Cognition. Internal processes of knowledge manipulation, including semantic analysis, decision making, etc.
3. Action. The control of muscle responses. 

This was originally intended as an explanation of complex human abilities of abstract problem solving, and was not meant to be a general theory of behavior. 

Indeed, in the brain, there is a great deal of parallelization and feedback, which complicates the picture.

These systems are not serial or independent.

• Perceptual Processing. The classical assumption of a unified and stable internal representation (an internal model of the external world) does not appear to be well supported. If there is something like a perception module, it overlaps so strongly with cognitive processes that the distinction between them becomes blurred.
• The representation of the external world generated by the visual system appears neither unified nor stable at the level of single neurons or neural populations. E.g., there are multiple representations of space in the parietal cortex. 
• This is further influenced by attentional modulation. The activity in visual regions is far from stable.
• Motor Control. Neural data do not support the assumption that by the time motor processing begins, cognitive processes have decided what to do, and only a single motor program is prepared before movement initiation.
• Many regions involved in movement planning are also active during movement execution.
• When planning activity has been studied in tasks where animals have choices, that same planning activity also appears related to decision making processes that should have been completed by the cognitive system. 
• Neural activity in motor regions appears to initially encode information about relevant stimuli and then changes to represent motor variables. 
• These additionally are modulated by a bunch of putatively cognitive variables. 
• Cognitive Functions. 
• Decision making does not appear to be localized within particular higher cognition centers, such as PFC. Instead, decisions (at least those reported through action) are made within the same sensorimotor circuits that are responsible for planning and executing the associated actions.

The distinctions among perceptual, cognitive, and motor systems may not reflect the natural categories of neural computations that underlie sensory-guided behavior. The framework of serial information processing may not be the optimal blueprint for the global functional architecture of the brain. 

An ecological perspective:

The hypothesis is that interaction with the environment involves continuous and simultaneous processes or sensorimotor control and action selection from among the distributed representations of a limited number of response options.