Embed Size (px)
Citation preview
Conclusions• The Lateral Occipital (LO) area encodes not only an object’s parts (Hayworth & Biederman, 2006) or local features (Op de Beeck et al, 2008), but also its medial axis structure.• LO is more sensitive to the medial axis structures of objects than to their global orientation, and this coding of axis structure is independent of local parts or features.• The coding of axis structure in LO is modulated by attention, but the response pattern in LO is not wholly determined by attention.
References (potential):
(1) Behrmann, M., Peterson, M. A., Moscovitch, M., & Suzuki, S. (2006). Independent representation of parts and the relations between them: evidence from integrative ag-nosia. Journal of Experimental Psychology: Human Perception and Performance, 32(5), 1169-1184.(2) Biederman, I. (1987). Recognition-by-components: a theory of human image under-standing. Psychological Review, 94(2), 115-147.(3) Connor, C. E., Gallant, J. L., Preddie, D. C., & Van Essen, D. C. (1996). Responses in area V4 depend on the spatial relationship between stimulus and attention. Journal of Neurophysiology, 75(3), 1306-1308.(4) David, S. V., Hayden, B. Y., Mazer, J. A., & Gallant, J. L. (2008). Attention to stimu-lus features shifts spectral tuning of V4 neurons during natural vision. Neuron, 59(3), 509-521.(5) Farah, M. (2004). Visual Agnosia (Second ed.). Cambridge, MA: MIT Press.Haxby, J. V., Gobbini, M. I., Furey, M. L., Ishai, A., Schouten, J. L., & Pietrini, P. (2001). Distributed and overlapping representations of faces and objects in ventral temporal cortex. Science, 293(5539), 2425-2430.(6) Hayworth, K. J., & Biederman, I. (2006). Neural evidence for intermediate represen-tations in object recognition. Vision Research, 46(23), 4024-4031.(7) Kriegeskorte, N., Mur, M., Ruff, D. A., Kiani, R., Bodurka, J., Esteky, H., et al. (2008). Matching categorical object representations in inferior temporal cortex of man and monkey. Neuron, 60(6), 1126-1141.
(8) Op de Beeck, H. P., Torfs, K., & Wagemans, J. (2008). Perceived shape similarity among unfamiliar objects and the organization of the human object vision pathway. Journal of Neuroscience, 28(40), 10111-10123.
References (used):
(1) Behrmann, M., Peterson, M. A., Moscovitch, M., & Suzuki, S. (2006). Independent representation of parts and the relations between them: evidence from integrative ag-nosia. Journal of Experimental Psychology: Human Perception and Performance, 32(5), 1169-1184. (2) Haxby, J. V., Gobbini, M. I., Furey, M. L., Ishai, A., Schouten, J. L., & Pietrini, P. (2001). Distributed and overlapping representations of faces and ob-jects in ventral temporal cortex. Science, 293(5539), 2425-2430. (3) Hayworth, K. J., & Biederman, I. (2006). Neural evidence for intermediate representations in object recog-nition. Vision Research, 46(23), 4024-4031. (4) Kriegeskorte, N., Mur, M., Ruff, D. A., Kiani, R., Bodurka, J., Esteky, H., et al. (2008). Matching categorical object representa-tions in inferior temporal cortex of man and monkey. Neuron, 60(6), 1126-1141. (5) Op de Beeck, H. P., Torfs, K., & Wagemans, J. (2008). Perceived shape similarity among unfamiliar objects and the organization of the human object vision pathway. Journal of Neuroscience, 28(40), 10111-10123.
MethodsSupport Vector Machine
Classifier Training• Stimuli were nine (3 objects x 3 views) or fifty-four (9 objects x 6 views) images of novel objects (see above).• Images appeared for 750 ms (3 object experiment) or 200 ms (9 object experi-ment) with an 8 second ISI. One image ap-peared on each trial.• MRI parameters: whole-brain scan of 2x2x2 mm voxels (3 object) or 2x2x2.5 mm voxels (9 object), 31 slices, TR = 2 s.
• Regions of interest were defined for each subject in independent lo-calizer scans.• For each region of interest, a support vector machine (SVM) classi-fier was trained on 7 runs of the data and tested on the 8th run, in order to test whether common axis structure and/or common body ori-entation produced consistently differentiable patterns in each region.
• Chance levels for classifi-cation accuracy were de-termined by classifying with random trial labels 100 times in each ROI.• To test different classifi-cation schemes against one another, correctly-labeled images were re-assigned to arbitrary categories (one of 280 possibilities for 9 images is shown above). A histogram of the 280 classifiers’ ac-
curacies in V1 and in LO is shown below to the left. • These distributions of classification accuracy were used to create z scores for the grouping schemes of interest.
Results:
Classify by Axis Structure
Classify by Cone Position
Classify by Body Orientation
Relative SVMClassification Accuracy
subj
ects
’ tas
k:
clas
sify
by
axis
st
ruct
ure
(n=
6)
subj
ects
’ tas
k:cl
assi
fy b
y co
ne
posi
tion
(n=
6)
Classify by Axis Structure
Classify by Component Parts
Classify by Body Orientation
*
subj
ects
’ tas
k:
clas
sify
by
axis
st
ruct
ure
(n=
6)
subj
ects
’ tas
k:cl
assi
fy b
y co
mpo
-ne
nt p
arts
(n=
6)
Relative SVMClassification Accuracy
Subject Performance Subject Performance
3 Objects x 3 Views Results: 9 Objects x
6 ViewsBackground
(Op de Beeck et al, 2008)
Presented at the Organization for Human Brain Mapping Conference, June 2010
Thanks to Jiye Kim, Xiaokun Xu, Ori Amir, Ken Hayworth, and Jonas KaplanSupported by NSF BCS 04-20794, 05-31177, 06-17699 to I.B.
References: (1) Behrmann, M., Peterson, M. A., Moscovitch, M., & Suzuki, S. (2006). Independent representation of parts and the relations between them: evidence from integrative agnosia. Journal of Experimental Psychology: Human Perception and Performance, 32(5), 1169-1184. (2) Haxby, J. V., Gobbini, M. I., Furey, M. L., Ishai, A., Schouten, J. L., & Pietrini, P. (2001). Distributed and overlapping representations of faces and objects in ventral temporal cortex. Science, 293(5539), 2425-2430. (3) Hayworth, K. J., & Biederman, I. (2006). Neural evidence for intermediate representa-tions in object recognition. Vision Research, 46(23), 4024-4031. (4) Kriegeskorte, N., Mur, M., Ruff, D. A., Kiani, R., Bodurka, J., Esteky, H., et al. (2008). Matching categorical object representations in inferior temporal cortex of man and monkey. Neuron, 60(6), 1126-1141. (5) Op de Beeck, H. P., Torfs, K., & Wagemans, J. (2008). Perceived shape similarity among unfamiliar objects and the organization of the human object vision pathway. Journal of Neuroscience, 28(40), 10111-10123.
• Object structure can be defined by specifying the relationships between the medial axes of an object’s parts.• Lesions to the ventral stream can produce an agnosia for object structure but leave intact the perception of the shapes of the indi-vidual parts (Behrmann et al, 2006).• Although there have been some human fMRI studies in-vestigating object representa-tions in ventral visual areas, these studies have not ad-dressed object structure spe-
cifically. With stimuli that vary in color, texture, and evolutionary signifi-cance (e.g., Haxby et al, 2001; Kriegeskorte et al., 2008), one cannot make inferences about
the role of object structure—or even object shape—in determining ventral stream response patterns.• PROBLEM: Could the BOLD signal in ventral visual areas dis-tinguish groups of objects that dif-fered only in their axis structures?Kriegeskorte et al 2008
(Behrman et al, 2006)
Lateral Occipital Cortex Represents Axis [email protected]://geon.usc.edu/~mark
Mark D. Lescroart & Irving BiedermanUniversity of Southern California Neuroscience Program
USC ImageUnderstanding Lab
