TY - GEN
T1 - Holistic processing is not always a property of right hemisphere processing- evidence from computational modeling of face recognition
AU - Galmar, Bruno
AU - Hsiao, Janet Hui Wen
PY - 2013
Y1 - 2013
N2 - The hemispheric asymmetry literature traditionally posits that holistic processing (HP) is a property of right hemisphere (RH) processing. Nevertheless, a counterexample was recently found: for Chinese character recognition expertise, studies showed reduced HP (as measured in the composite task) and increased RH lateralization, revealing that these two effects may be separate processes. With a computational model of face recognition, in which we implement a theory of hemispheric asymmetry in perception that posits a low spatial frequency bias in the RH and a high spatial frequency bias in the left hemisphere (i.e., the Double Filtering by Frequency Theory of Ivry and Robertson), here we show that when the face recognition task relies purely on featural information, there is a negative correlation between HP and RH lateralization: HP increases whereas RH lateralization decreases with increasing stimulus dissimilarity. In contrast, when the face recognition task relies purely on configural information, there is a strong positive correlation between HP and RH lateralization: both HP and RH lateralization increase with increasing stimulus dissimilarity. These results suggest that HP and RH lateralization are separate processes that can be influenced differentially by task requirements.
AB - The hemispheric asymmetry literature traditionally posits that holistic processing (HP) is a property of right hemisphere (RH) processing. Nevertheless, a counterexample was recently found: for Chinese character recognition expertise, studies showed reduced HP (as measured in the composite task) and increased RH lateralization, revealing that these two effects may be separate processes. With a computational model of face recognition, in which we implement a theory of hemispheric asymmetry in perception that posits a low spatial frequency bias in the RH and a high spatial frequency bias in the left hemisphere (i.e., the Double Filtering by Frequency Theory of Ivry and Robertson), here we show that when the face recognition task relies purely on featural information, there is a negative correlation between HP and RH lateralization: HP increases whereas RH lateralization decreases with increasing stimulus dissimilarity. In contrast, when the face recognition task relies purely on configural information, there is a strong positive correlation between HP and RH lateralization: both HP and RH lateralization increase with increasing stimulus dissimilarity. These results suggest that HP and RH lateralization are separate processes that can be influenced differentially by task requirements.
KW - Connectionist modeling
KW - Face processing
KW - Hemispheric lateralization
KW - Holistic processing
UR - http://www.scopus.com/inward/record.url?scp=84893344690&partnerID=8YFLogxK
U2 - 10.1007/978-3-642-42054-2_1
DO - 10.1007/978-3-642-42054-2_1
M3 - 會議論文篇章
AN - SCOPUS:84893344690
SN - 9783642420535
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 1
EP - 8
BT - Neural Information Processing - 20th International Conference, ICONIP 2013, Proceedings
T2 - 20th International Conference on Neural Information Processing, ICONIP 2013
Y2 - 3 November 2013 through 7 November 2013
ER -