R Q MBC
<p class="MsoNormal" >RODRIGUEZ, Christian, Intertemporal Choices and Values, Mar 7, 2014
This is a 3:15PM FriSem of MBC Center in Jordan Hall, Stfd.</p>
<p>
he shows his old car that he loved but he faced a big mechanic bill;
so, do you fix the car or buy a new one?</p>
<p>
<b>do you spend now or save money for a rainy day</b>.</p>
<p>
where are and how are discounted values encoded?
(he is in <b>Sam McClure's</b> Lab)</p>
<p>
2006 Padoa-Schioppa: OFC recording: = orbitofrontal cortex
encodes rewards like 1 drop of grape juice vs 3 drops of water..
monkey data
<b>some neurons track the value of the offer = Offer value</b></p>
<p>
versus: <b>neurons that track the chosen value</b>... (the actual outcome of the decision)
bottom line of their research: <b> the OFC does it all.</b></p>
<p>
vs moving dots type expt
as in Gold and Shadlen in 2007:
see integration of evidence (a la Ratcliff)
differential rates of increase of evidence toward a decision.</p>
<p>
Hare 2011</p>
<p>
dmPFC dorsomedial PFC
fit a drift diffusion model...
and pPC... posterior parietal cortex.</p>
<p>
but their monkeys just making decisions abt current choices..</p>
<p>
are these regions true integration regions.</p>
<p>
<b>OUTLINE for today</b>:</p>
<p>
<b>vmPFC and OFC encode temporal discounted values </b>...</p>
<p>
next, <b>discounted value accumulation accounts for intertemp choice behavior</b>...</p>
<p>
value accumulation occurs in vmFC...</p>
<p>
experiment = expt: 10 bucks now vs more in future:</p>
<p>
( eg 19 bucks in 30 days...)</p>
<p>
$30 presented for 1000 ms.</p>
<p>
and 19 bucks displayed for 4000 ms.</p>
<p>
then do either EEG or fMRI</p>
<p>
value discounted Vd = r/ 1 + kt (hyperbolic discount)</p>
<p>
prob of a reward is either .1, .3 .5 .7 or .9</p>
<p>
where is value encoded?</p>
<p>
Value of delayed reward:</p>
<p>
R OFC stores the discounted value</p>
<p>
and in vmPFC for the Net value...</p>
<p>
but what is encoded in OFC is it</p>
<p>
Vd... Vd + Vi or is it
Vd - Vi..</p>
<p>
he looks at group averages....</p>
<p>
Anthony: exact location of region matters...
large diff btwn ventral surface and interior...</p>
<p>
so... we DO have tempr discounted value encoded in vmPFC and OFC...</p>
<p>
next point:
does discounted value accumulation account for intertemproal choice behavior...</p>
<p>
(he fits a k for each subject...
may reflect a v impulsive subject vs delayed gratification)</p>
<p>
*</p>
<p>
LBA linear ballistic accumulation
Value accum model... as in Ratcliff...</p>
<p>
this is from Brown and Heathcote in 2008</p>
<p>
2 accumulators: one for immed reward and one for the delayed reward...</p>
<p>
evidence is gradualy accumulating for each .</p>
<p>
main param is the drift rate</p>
<p>
tau is just time to press the button (time that is irrelevant to the decision)</p>
<p>
he shows RTs for difft probabilities</p>
<p>
eg is prob is only 0.1 of getting the delayed reward, then S goes for the immed reward, etc.</p>
<p>
he showed that form of discounting model did not matter (whether hyperbolic or exponential)</p>
<p>
next point:</p>
<p>
value accumulation localizes in dmFC frontal cortex</p>
<p>
can we explain trial specific data...</p>
<p>
want to estimate a start point and a drift rate to account for response time for each trial.</p>
<p>
use simulated data...</p>
<p>
TVA = total value accumulation</p>
<p>
dm FC = dorsomedial frontal cortex was only region that survived GLM.</p>
<p>
(this is the only region that is similar to LIP in visual motion literature)</p>
<p>
</p>
<p>
last point:</p>
<p>
related but dissociable functions localized in pPC and latPFC</p>
<p>
funcitonal connectivy in neural oscillations...</p>
<p>
PLVs = phase locking values measure dynamic episodes of synch oscillation...</p>
<p>
where do see evid. of func connectivity..</p>
<p>
lPFC = lateral PFC...</p>
<p>
seems that pPC = poster parietal cortex is mainly related to the display of the amount.</p>
<p>
pPC is position related to increased delay... but not known why...</p>
<p>
</p>
<p>
</p>