Brain Activity and Complex Motion

1
Brain Activity and Complex Motion

• The usual way of studying brain activities is in
  a well controlled environment with ability to
  repeat each condition many times.
• In motor physiology this is very often done by
  studying single unit activity in relation to a
  reaching movement.
• What can we see with more natural motion?
• 1. Scribbling
• 2. Prehension

2

• Do motor cortical neurons change their preferred
  directions when switching between different motor
  tasks ?

3
CenterOut (CO) task and trajectories
4
An example of a scribbling trajectory
5
Extracting CO like segments out of the
scribbling trajectory.

• Velocity profile was segmented between adjacent
  minima
• Segments were cleaned based on duration from
  start to peak velocity, average direction and STD
  of instantaneous direction within segment.

6
The final scribbling and CO trajectories
7
Similar Directional tuning during CO and
scribbling movements
8
Different directional tuning during CO and
scribbling movements
9
PDs during CO and scribbling tended to be
different

• In 13 out of 20 cells (65) PDs during
  scribbling were significantly different than PDs
  during reaching movements.

10
Possible causes for the difference in PDs

• Other movement parameters change between
  scribbling and CO
• tangential velocity
• Initial position
• curvature of the segments
• tangential acceleration
• The differences are task related

11
Differences in PDs were not explained by
variability in other movement parameters

• Scribbling segments were divided based on high
  and low values of different movement parameters.
• PDs were obtained for each subgroup separately
• Results from PDs comparison
• No significantly different PDs due to variability
  in peak velocity and peak acceleration.
• 1/13 (8) cells with significantly different PDs
  due to variability in segments directional STD
• 2/13 (15) cells with significantly different PDs
  due to variability in initial position.

12
The Free Tracing (FT) task trajectories
13
PDs during FT and CO movements tended to be
similar

• Lower differences of PDs between CO and FT
  relative to CO and scribbling PD differences.
• Only 10 out of 49 cells (20) had significantly
  different PDs during FT and CO movements.

14
Summary and conclusions

• Directionally tuned cells during both CO and
  scribbling movements tended to have different
  preferred directions during each type of
  movement.
• These differences were not explained by the
  variability in various movement parameters.
• These differences were less frequent when the
  monkey alternated between CO and FT tasks.
• Therefore directional tuning of motor cortical
  cells is not only movement but also task
  dependent.

15
Brain Activity and Complex Motion

• Most of what we do or perceive is compositional.
• We compose sounds into phonemes phonemes into
  words words into sentences
• What are the neuronal correlates of these
  properties?
• In scribbling what happens when two pieces of
  motion are concatenated?

16
Concatenating Movements (Tishby Gat)

• Compute the likelihood of change in firing rate
  for all cells.

17
Concatenating Movements

• Compute the likelihood of change in firing rate
  for every cell.
• Find which cells tend to change their firing
  rates just before
• start of movement.

18
Concatenating Movements

• Compute the likelihood of change in firing rate
  for every cell.
• Find which cells tend to change their firing
  rates just before
• start of movement.
• peak tangential velocity.

19
Concatenating Movements

• Compute the likelihood of change in firing rate
  for every cell.
• Find which cells tend to change their firing
  rates just before
• start of movement.
• peak tangential velocity.
• trough in .

20
Concatenating Movements

• BUT cell assemblies overlap.
• One needs to know who is firing AND who is quiet.
• Problem with low firing rates and sparse sampling.

21
Prehension

• It is still unclear what happens neuronally when
  1 element of motion is concatenated to another.
• Compositionality can manifest itself also by
  combining elements in parallel (like lines to a
  figure). In motor systems that happens, for
  instance, during prehension
• We can pick any object from any location. Thus,
  all combinations of grasping and reaching may be
  combined.

22
Field Potential Oscillations in Posterior
Parietal Cortex During Reaching and Grasping
Movements
23
Reaching Grasping are Mediated by Separate
Parieto-Premotor Channels (Kandel, Schwartz
Jessell, 4th Ed.)
Reaching MIP MDP (Andersen), Area 5
(Kalaska) PMdc (Wise, Kalaska) Grasping -
AIP (Sakata) F5 in PMv (Rizzolatti) Unit
properties Directional Tuning Object
Specificity Bidirectional, Segregated Connection
s
24
Objectives

• Train monkeys to reach grasp various objects in
  various directions.
• Record simultaneously from Reaching- related
  and Grasping-related areas.
• Search for signs/mechanisms of inter-area
  coordination.
• This talk focus on LFP oscillations, which were
  suggested as a binding mechanism for distributed
  representations (Singer Gray, 1995)

25
Task Setup Protocol

• Touch pad in center of workspace
• 6 target locations X 3 prehension objects
• Controlled Sound Light conditions
• Epochs Control, Signal, Set, Pre-Go, RT-MT,
  Hold.

movie
26
(No Transcript)
27
Prehension objects
Plate Finger opposition
Box Power grip
Precision grip object
Reaching pad
28
movie
29
Time Domain LFP traces show task dependent
modulation, including oscillations
30
Frequency domain time resolved spectrum shows
epoch-dependent changes in spectral composition
Alpha 8-13 Hz Beta 13-30 Hz Gamma 30-60 Hz
31
Beta oscillations in SPL show directional
selectivity, with non-uniform PD distribution
32
This is very different from tuning of MU spikes
in the same area
33
Great expectations

• Non-Uniform tuning distribution exists Both in
  Oscillations and in RMS of signal.
• This is consistent with Motor Cortex results
  (Donchin et al., 2001).
• We are ready to look at between-area effects
  (Coherence).
• BUT

34
Problem Typical AIP data do not show beta
oscillations (may show gamma oscillations)
Alpha 8-13 Hz Beta 13-30 Hz Gamma 30-60 Hz
compare
35
Within and between area coherence a
measure of coordination?
d.78mm
d1.53mm
d14.77mm
36
Between-Area coherograms
37
Significant coherence is related (time
frequency-wise) to Evoked Potential phenomena,
not beta/gamma oscillations
38
Summary

• Beta Oscillations very frequent in SPL, Gamma
  Oscillations are less frequent, more in IPL.
• Our results do not comply with previously
  suggested explanations / functions of
  oscillations
• (1) Fast oscillations are signs of focused
  attention states (Murthy Fetz, 1996). This is
  the reverse of SWS.
• (2) Motor cortex beta oscillations are useful
  for efficient motor output state, in contrast to
  high processing capacity (Baker et al., 1999).
  
• (3) Gamma oscillations serve to bind
  distributed cortical representations (Singer
  Gray, 1995)

39
Some neural mechanisms of cortico-cortical
cooperation
40

• Question
• Do, and how do, cortical areas coordinate their
  activity
• Model system
• PM (pre-motor) cortex
• Dorsal PM reaching-related (Kalaska, Wise)
• Ventral PM grasping-related (Rizzolatti)

41
Temporal coordination hypothesis

• Crosstalk between areas
• At behaviorally relevant time scales
• Modulated by context
• Tests
• Local field potential pair-wise correlations
• Single unit cross-correlations

42
LFP pair-wise correlationsthe raw data
PREGO
RTMT
43
Zero-lag modulation by distance
44
Exponential decay with distance
45
Binned by distance
46
Modulation by behavior
47
Exponents coefficients differ
48
Significance

• 2 way ANOVA
• distance
• epoch
• interaction ns
• Rank test
• lt 0.05
• lt 0.01
• lt 0.001

Control Signal Set Prego Rtmt Hold
Control - ns ns ns
Signal - - ns ns
Set - - - ns ns
Prego - - - -
Hold - - - - -
49
Spike-to-spike cross-correlationsthe raw data
GO signal
Cue On
50
Zoom in
51
Same area, different electrodes
52
Same area, different electrodes
53
Same area, different electrodes
54
Cooperation during preparation
Cue Off
GO signal
55
CC across areas
56
Movement specific cooperation
Movement initiation
Movement termination
57
Longer lags between areas
58
Statistics differ
59
Summary

• Both local field potentials and single units seem
  to coordinate their activity across distances in
  a precise, context-related manner
• Temporal coordination hypothesis supported

60
Brain Activity and Complex Motion

• Conclusion
• Life is tough

61
(No Transcript)
62
Extra Figures
63
Power in epochs (lumped) MIP
RMS PWR tot Alpha Beta Gamma Other
64
Power in epochs (lumped) AIP
RMS PWR tot Alpha Beta Gamma Other