This preview shows page 1. Sign up to view the full content.
Unformatted text preview: nt (this quantity will tend to
overestimate the evidence from weak stimuli and underestimate the evidence from strong stimuli). Accordingly,
the evidence that accumulates during a trial can be
interpreted as a fraction of this quantity and thus in units
of natural bans—even when the scaling between the
decision variable and the weight of evidence is not
known (e.g., if the brain does not know the shapes of
the sensory response distributions).
1 is a constant that represents the barrier height
f h . For two mutually exclusive hypotheses
1
Pr(h m) or, equivalently, Pr(h )
1
nd assuming the weight of evidence is exin bans, this rearranges to:
Review
303 Pr(h m) 1 1
.
10 (4) LIP neurons?
4 indicates that the posterior probability of h 1 only on the value of the barrier, B, and not on
ular samples of evidence, m, encountered. In
rds, as long as the weight of evidence reaches
he probability that h is correct is a fixed value.
Figure 2. Decision Model 1 logarithm of the ratio of the
equal to the
between x and y. Solving for the weight of evidence (in
nically related
obabilities. Incommon variance,
this case, for a given bans)are the versus h yields:
weightfor h means of
2
is the
units of natural
1966). Briefly,
1 and
0
1
0
(e.g., the stopping pointExperimentalsEvidence
in Banburismus): the covariance
Turing’s and y, of
x Weight respectively, under h1, and i
is case, a rate Evidence
Pr(x For
Neurophysiology: Banburismus h ) the Brain
s from a pair of sensory neurons whenfor or h1 is true, asof evidence 1(in
h0evidence
indicated.,y in
between x
Solving
thought of as and y. weight of the weight log
Pr(h1m) neuron that prefers upward motion. The lower curve would
tionselective ofweight of evidence versus h (3)
units
natural bans) for h1 B
0 yields:Pr(x,yh0)
lushconditions
Pr( 0m)
fers downward motion. Here we make the simplifying assumption that the
value but can
d equal variances ( 2 ). Note that other forms x,yh distribution, including
of the )
1
(1
Pr(
0)
1
·
· (x y) (6)
adlen, 2001). thatpanel on the rightlog
shows Turing’s weight
constant The...
View
Full
Document
This note was uploaded on 09/15/2011 for the course COGS 1 taught by Professor Lewis during the Spring '08 term at UCSD.
 Spring '08
 LEWIS

Click to edit the document details