Title: The temporal extent of holistic processing
11
The temporal extent of holistic processing Jed
Singer and David Sheinberg
22
What is a composition?
Some things?
An arrangement of things?
A new thing unto itself?
33
The Composite Face Effect (CFE)
It is more difficult to identify one half of a
face when the other half has a different identity
This phenomenon is reduced when the pieces are
separated or the face is inverted
44
Classical results (RT differences)
Inversion eliminates interference from
properly-aligned bottom half in recognition task
(Young et al. 1987)
55
The CFE as an indicator of holistic processing?
One line of reasoning The composite face effect
shows that we can not disregard distracting parts
of a face when they are presented in a normal
configuration with the attended parts of a face
because we see faces as immediate and
undifferentiated wholes. An alternative The
percept of a particular face identity at a
particular location arises from an accumulation
of evidence (for and against) this accumulation
might even span spatial or temporal disjunction.
66
Gauthier, Tanaka, Brown (submitted)
Subjects study only one of two dissociated face
halves Yet, test response (same/different) is
delayed in aligned trials when the ignored half
would indicate a different response than the
studied half
77
Does the CFE survive temporal discontinuity?
Four faces each of Clinton and Kerry, for 32
chimeras 8 originals Original images repeated
four times each orientation 11 SOAs, upright and
inverted 1408 total trials Noise matches average
spatial spectrum of faces
88
Does the CFE survive temporal discontinuity?
Subjects report identity of interior face,
instructed to focus on accuracy we expect to see
a composite face effect at 0 SOA, but will we see
it for asynchronous presentations?
real-time example
slow example
99
Splitting faces across time is no big deal
whole faces
split faces
upright
inverted
upright chimeric
inverted chimeric
1010
Our CFE measure (correct reaction times)
-
interference
-
interference when inverted
-
CFE magnitude
1111
Zero-lag (classical) CFE replicated
-
229 ms
764 ms
-
535 ms
-
155 ms
819 ms
664 ms
74 ms
1212
CFE results!
CFE magnitude (ms)
permutation test p-values
.05
.01
N 11
.001
exterior-interior SOA (ms)
1313
In conclusion
Replicate classical composite face effect Effect
is even stronger when distracter part follows Are
well-known faces processed more holistically
when their parts are presented sequentially? Hol
istic processing is a complicated issue!