Occam modelling

Used information is minimal
makes the least possible assumptions
considers those as „axioms” (facts, starting points)
derives every other property

Starting assumptions

1st assumption:
controller and neocortex have functional similarities

2nd assumption:
start with controller

Features

Control deals with directions
it could deal with trajectories in
configurational space
external space

Limb moves along straight lines

Corollary: SFT

SFT = Speed Field tracking
TT = Trajectory tracking

Feature

There is a differencing system in the brain
We assume that:
every difference is computed in the BG, e.g.:
acceleration = difference of speeds
speed = difference of coordinates

Imprecise inverse dynamics

ID = inverse dynamics

move mouse over picture!

Similarity

Predictions

Long and tunable delays in DG loops
OK -> confirmed by the Buzsáki group (Henze et al., Nature Neuroscience. 5: 790–795, 2002)

Persistent activities in EC deep layers but not in EC superficial layers
OK -> confirmed by the Hasselmo group (Egorov et al. Nature, 420: 173–178, 2002)

LTM: Feedback between areas
OK -> recurrent interactions are necessary for visual awareness (Lamme TICS 7:12-18, 2003)
BUT: why so quiet?