Abstract-- Lateralization of perception of various facial attributes (age, attractiveness, gender, lip-reading and expression) was studied using chimaeric faces in which the sides of the face differed along one dimension (e.g. the left side was male and the right side female). Computer graphics were used to eliminate naturally occurring physical asymmetries (e.g. those present in the mouth during speech and spontaneous smiles) and obvious vertical mid-line joins in the photo-realistic chimaeric stimuli. Following previous studies, we found that subjects ' judgements of gender and expression were influenced more by the left than the right side of the face (viewer’s perspective). This left of face stimulus bias extended to judgements about facial attractiveness and facial age. This was not true of lip-reading stimuli; for these stimuli subjects were influenced more by the right than the left side of the face. Thus using free fixation, it appears possible to demonstrate in normal subjects that brain processes underlying judgements of facial speech display different lateralization from the judgements of other facial dimensions.

Address for correspondence:

In visual word recognition, it is well-known that there exist preferred landing positions for eye fixations. However, the existence of preferred landing positions in face recognition is less well established. It is also unknown how many fixations are required to recognize a face. To investigate these questions, we recorded eye movements during face recognition. Subjects were allowed a variable number of fixations before the stimulus was masked during an otherwise standard face recognition task. We find that optimal recognition performance is achieved with two fixations – performance does not improve with additional fixations. The first fixation is just to the left of the center of nose, and the second on the center of the nose, suggesting preferred landing positions for face recognition. Furthermore, the fixations made during face learning are different in location from and more variable in duration than at face recognition time, suggesting that different strategies are used. 2

Hemispheric asymmetries in the perception of local and global features have been consistently reported: there is an advantage for responses to global features in the left visual field/right hemisphere and an advantage for responses to local features in the right visual field/left hemisphere. It has been proposed that this asymmetry originates from differential frequency bias in the two hemispheres (e.g., Ivry & Robertson, 1998). Nevertheless, there is little evidence supporting hemispheric specialization for particular frequency ranges (e.g., Fendrich & Gazzaniga, 1990) or differential frequency tuning in the neurons in the two hemispheres. Here we test the hypothesis that this hemispheric asymmetry in visual perception takes place at the encoding stage beyond the sensory level. We use two autoencoder networks with differential connectivity configurations as the way to develop differential encoding in the two hemispheres, to reflect the anatomical evidence that there is more interconnectivity among the neighboring cortical columns in the right hemisphere than the left hemisphere (e.g. Hutsler & Galuske, 2003). We show that this differential encoding mechanism has a better fit with human data than the model based on differential frequency bias, and thus is a more anatomically realistic and cognitively plausible model in accounting for the hemispheric asymmetry in visual perception.

Visual word recognition in alphabetic languages such as English has been shown to have left hemisphere (LH) lateralization and argued to be linked to the LH superiority in language processing. Nevertheless, Chinese character recognition has been shown to be more bilateral or right hemisphere (RH) lateralized and thus is a counterexample of this claim. LH processing has been shown to have a high spatial frequency (HSF) bias, whereas RH processing has a low spatial frequency bias. Through computational modeling, here we test the hypothesis that English word recognition is lateralized to the LH and Chinese to the RH due to visual characteristics of words instead of language lateralization. We show that at least two factors may account for this dichotomy: (1) Visual similarity among words: The smaller the alphabet size is, the more similar the words in the lexicon are, and the more the model relies on HSFs to distinguish words. (2) The requirement to decompose words into letters in order to map to phonemes during learning to read English: Mapping word input to its constituent letters requires more HSF information compared with mapping to its word identity. English has a large lexicon size but only 26 letters, whereas Chinese has a much smaller lexicon with a much larger “alphabet ” (stroke patterns). In addition, Chinese is a logographic system: stroke patterns do not map to phonemes and thus no decomposition is required. Hence, the lateralization of visual word recognition in different languages may depend on visual characteristics of words instead of the LH language lateralization as previously thought.

When human subjects view photographs of faces, their judgments of identity, gender, emotion, age and attractiveness depend more on one side of the face than the other. We report an experiment testing whether allocentric kin recognition (the ability to judge the degree of kinship between individuals other than the observer) is also lateralized. One hundred and twenty four observers judged whether or not pairs of children were biological siblings by looking at photographs of their faces. In three separate conditions (1) the right hemi-face was masked, (2) the left hemi-face was masked, or (3) the face was fully visible. The d � measures for the masked left hemi-face and masked right hemiface were 1.024 and 1.004 respectively (no significant difference) and the d � measure for the unmasked face was 1.079, not significantly greater than that for either of the masked conditions. We conclude, first, that there is no superiority of one or the other side of the observed face in kin recognition, second, that the information present in the left and right hemi-face relevant to recognizing kin is completely redundant, and last that symmetry cues are not used for kin recognition. [188 words] 3

exploration of the relationship between holistic processing and right

In Chinese orthography, a dominant structure exists in which a semantic radical appears on the left and a phonetic radical on the right (SP characters); the minority, opposite arrangement also exists (PS characters). Recent studies showed that SP character processing is more left hemisphere (LH) lateralized than PS character processing; nevertheless, it remains unclear whether this is due to phonetic radical position or character type frequency. Through computational modeling with artificial lexicons, in which we implement a theory of hemispheric asymmetry in perception but do not assume phonological processing being LH lateralized, we show that the difference in character type frequency alone is sufficient to exhibit the effect that the dominant type has a stronger LH lateralization than the minority type. This effect is due to higher visual similarity among characters in the dominant type than the minority type, demonstrating the modulation of word type frequency on hemispheric lateralization.