Neural Synchrony and Selective Attention
<p>Dec 08, 2013 03:12 PM</p>
<p>
video lecture by Bob Desimone, given 2010 at Boston Univ...
this is the keynote... a President's Lecture in Neurosci at the BU Center for Neurosci.</p>
<p>
intro: Petrucchio: it is the mind that makes the body rich (from Taming of the Shrew);
(intro by Bob Brown; Pres of BU (chem Engr by training; was at MIT for 25 yrs
before moving to BU as 10th pres of BU))</p>
<p>
Howard Eichenbaum introduces Bob D.</p>
<p>
(Bob was a grad student with Charlie Gross;
later was Lab Direc at NIMH Intramural Program)</p>
<p>
Direc of McGovern at MIT;</p>
<p>
how brain directs attn;</p>
<p>
Center for Cognit and Neuro Systems:
he singles out Nancy Koepel.</p>
<p>
SELECTIVE ATTENTION </p>
<p>
shows a student in a crowded room, looking at a book</p>
<p>
shows change blindness: do not pick up major changes...
only aware of a tiny bit of the scene in front of us.</p>
<p>
studies visual pathway of monkey</p>
<p>
V1 to V2 to V4 to TEO to TE
PFC connects to V4 and to TEO
Parietal separ connects....</p>
<p>
John Reynolds worked with Steve Gr. and Gail at BU</p>
<p>
see Reynolds, Chelazzi Desimone in 1999;</p>
<p>
key idea: selective attn works at least in part by temporal synchrony;</p>
<p>
<b>thousands of inputs but only ONE OUTPUT</b> !!!
<b>neurons are like us in MICROCOSM </b>!!!! what is important!!!</p>
<p>
his KEY SLIDE is at 14:40 (min) labeled Temporal Synchrony</p>
<p>
"how can attention incr the effectieness of neural pathways for behaviorally rlevant stim
at the expnese of distractors? Answer is Temporal Synchrony.</p>
<p>
How does top-down attentional bias change aciton potential trains:</p>
<p>
what controls the output of a neuron:</p>
<p>
if inputs are synced then more likely to exert influence and cause firing
(cuz of limited integraito times)</p>
<p>
inputs need to be synched within a few ms.</p>
<p>
fromb Bing and Poo in 98. can get LTP
but if dissoc then get LT Depres.</p>
<p>
HC cell types sync at difft phases of cycle in theta;
can carries impt info abt where animal is in space.</p>
<p>
How much does synchrony matter?</p>
<p>
anatomy of synchrony? eg by lamina in V4: dramatic diff. across layers:
gamma synch only occurs in superficial layers of V4 (incr gamma and decr alpha)
(deep layers (lamina) of cortex do NOT show this)</p>
<p>
attentional effects on firing rates is mainly in (? only in) superfic layers;</p>
<p>
moves backwards (feedback);
superficial layers are the ones that communic cortex to cortex
(deep layers communic with subcortex)</p>
<p>
Steve Grossberg thinks that superfic layers are controlling expectation.</p>
<p>
attn seems to work vis top-down feedback from upstream areas toward earlies areas.</p>
<p>
top down from both Parietal area and from FEF... problems with top-down control of attn.</p>
<p>
FEF controls V4... did simult recording in both of those areas.</p>
<p>
collab with Steve Gott;</p>
<p>
attn enhancing synchrony btwn V4 and FEF...
with a phase shift of abt 10 ms. (abt half a cycle)</p>
<p>
always takes about 10 ms (irrespective of freq)</p>
<p>
takes time for neural signals to be propagated... brain MUST take into account the time delays;
must take propation delays into account (therefore cannot have zero phase delay)</p>
<p>
therefore, to get self-reinforcing oscillations must take into acct propag. delays.</p>
<p>
FEF initiates these effects and then will be self-reinforcing...</p>
<p>
this is a genl principle:</p>
<p>
Vinnie Soder in Australia see Saalmann in Science 2007
btwn MT and LIP</p>
<p>
and Earl Miller see Buschmann in Science;</p>
<p>
human subjects using MEG: look at ms time scale in humans (but poor spatial res)
similar results to monkeys;</p>
<p>
get coherent activity;</p>
<p>
</p>
<p>
does synchrony play a role in visual search for target features?</p>
<p>
(normally we look around the room and attend to things under our own control...)</p>
<p>
visual search (like Where's Wald0)</p>
<p>
find the flag within big array os stim...</p>
<p>
if stim in Rec Field matches the cue then gamma sync increases.</p>
<p>
when animal has selected the target... then you get maximum synch...</p>
<p>
do combined recording from V4 and FEF...</p>
<p>
cells are saying.... "we have what you are looking for!" (even before the saccade to the location)</p>
<p>
set up bias maps that is input to FEF;</p>
<p>
go from feature maps in visual areas to cause selection in FEF;</p>
<p>
Pascal Fries in Nijmegen.</p>
<p>
</p>
<p>
</p>
<p>
</p>
<p>
</p>
<p>
</p>
<p>
</p>
<p>
</p>
<p>
</p>
<p>
</p>
<p>
</p>
<p>
</p>
<p>
</p>
<p>
</p>