After arriving in the primary visual cortex V1, there are two streams (dorsal (where/how) and ventral (what)) to grasp the visual.
Optical ataxia + visual agnosia are two conditions that “prove” the existence these streams.
How obtained from single-cell data?
Dorsal stream → AIP (anterior intraparietal sulcus) is the grasping area!
Ventral stream → LOC (lateral occipital complex) via V4 (texture) is the object recognition area!
Codes for object identity (almost independent from shading, illumination, …)
PMv (ventral premotor complex) has a dictionary of motor actions.
Mirror neurons = fire when entity is performing a specific action, but also when the entity is observer to some other entity performing a (known) action.
General - How was it discovered that these different areas are responsible for relatively “fixed” functions (e.g. V3 for color/shading)?
Owl monkey brain is easier since it has less folds (= less complexity in terms of pathways).
(Speculative) → Cortical hierarchy is a closed-loop system where sensory input has effect on the motility and vice versa.
Frontal cortex neurons modulate firing → Activity rises until go signal, then decreases after motor action has been completed.
After some time, more neurons become tuned to the directions (i.e. learning!)
Motor Goal Encoding & Reach Movement Planning in PRR & PMd
Space-context integration for reaching → a positive feedback loop happens between the PRR and PMd, both becoming sensorimotor. When this loop has reached a certain threshold, branching to the motor area happens.
After obtaining a radial tuning curve, we could in theory train a machine-learning decoder model to predict in which direction the monkey will move their arm. As can be seen from the curve, after a cue is given, the decoder can predict (> 90% accuracy) which direction the monkey will move their arm in.
The 25% chance level is a result of the axis of movement, not the direction (8 possible directions = 4 possible axis of movement).
Goal-directed Movement Planning
spatial constraints + behavioral context ⇒ goal-directed movement plan
Anti-reach → Show different color spot → monkey learns to reach to the opposite side of the movement axis!
For a decoder, if we base ourselves on the visual cue only, we can not properly predict the spatial direction of movement. If we focus on the motor-goal, then we can!
Visually-guided Grasping
From eye movement of monkey, prediction of grasping location can be made.
