Re-evaluating split-fovea processing in word recognition: Effects of fixation location within words

Investigating the Effectiveness of Spatial Frequencies to the Left and Right of Central Vision during Reading: Evidence from Reading Times and Eye Movements

Printed words are complex visual stimuli containing a range of different spatial frequencies, and several studies have suggested that various spatial frequencies are effective for skilled adult reading. But while it is well known that the area of text from which information is acquired during reading extends to the left and right of each fixation, the effectiveness of spatial frequencies falling each side of fixation has yet to be determined. To investigate this issue, we used a spatial frequency adaptation of the gaze-contingent moving-window paradigm in which sentences were shown to skilled adult readers either entirely as normal or filtered to contain only low, medium, or high spatial frequencies except for a window of normal text around each point of fixation. Windows replaced filtered text either symmetrically 1 character to the left and right of each fixated character, or asymmetrically, 1 character to the left and 7 or 13 to the right, or 1 character to the right and 7 or 13 to the left. Reading times and eye-movement measures showed that reading performance for sentences presented entirely as normal generally changed very little with filtered displays when windows extended to the right but was often disrupted when windows extended to the left. However, asymmetrical windows affected performance on both sides of fixation. Indeed, increasing the leftward extent of windows from 7 to 13 characters produced decreases in both reading times and fixation durations, suggesting that reading was influenced by the spatial frequency content of leftward areas of text some considerable distance from fixation. Overall, the findings show that while a range of different spatial frequencies can be used by skilled adult readers, the effectiveness of spatial frequencies differs for text on each side of central vision, and may reflect different roles played by these two areas of text during reading.

Aging and the optimal viewing position effect in visual word recognition: Evidence from English

Words are recognized most efficiently by young adults when fixated at an optimal viewing position (OVP), which for English is between a word’s beginning and middle letters. How this OVP effect changes with age is unknown but may differ for older adults due to visual declines in later life. Accordingly, a lexical decision experiment was conducted in which short (5-letter) and long (9-letter) words were fixated at various letter positions. The older adults produced slower responses. But, crucially, effects of fixation location for each word-length did not differ substantially across age groups, indicating that OVP effects are preserved in older age.

Effects of word length on eye movement control: The evidence from Arabic

The finding that word length plays a fundamental role in determining where and for how long readers fixate within a line of text has been central to the development of sophisticated models of eye movement control. However, research in this area is dominated by the use of Latinate languages (e.g., English, French, German), and little is known about eye movement control for alphabetic languages with very different visual characteristics. To address this issue, the present experiment undertook a novel investigation of the influence of word length on eye movement behavior when reading Arabic. Arabic is an alphabetic language that not only is read from right to left but has visual characteristics fundamentally different from Latinate languages, and so is ideally suited to testing the generality of mechanisms of eye movement control. The findings reveal that readers were more likely to fixate and refixate longer words, and also that longer words tended to be fixated for longer. In addition, word length influenced the landing positions of initial fixations on words, with the effect that readers fixated the center of short words and fixated closer to the beginning letters for longer words, and the location of landing positions affected both the duration of the first fixation and probability of refixating the word. The indication now, therefore, is that effects of word length are a widespread and fundamental component of reading and play a central role in guiding eye-movement behavior across a range of very different alphabetic systems.

Aging and the optimal viewing position effect in Chinese

Substantial evidence indicates that where readers fixate within a word affects the efficiency with which that word is recognized. Indeed, words in alphabetic languages (e.g., English, French) are recognized most efficiently when fixated at their optimal viewing position (OVP), which is near the word center. However, little is known about the effects of fixation location on word recognition in non-alphabetic languages, such as Chinese. Moreover, studies to date have not investigated if effects of fixation location vary across adult age-groups, although it is well-established that older readers experience greater difficulty recognizing words due to visual and cognitive declines. Accordingly, the present research examined OVP effects by young and older adult readers when recognizing Chinese words presented in isolation. Most words in Chinese are formed from two or more logograms called characters and so the present experiment investigated the influence of fixation location on the recognition of 2-, 3-, and 4-character words (and nonwords). The older adults experienced generally greater word recognition difficulty. But whereas the young adults recognized words most efficiently when initially fixating the first character of 2-character words and second character of 3- and 4-character words, the older adults recognized words most efficiently when initially fixating the first character for words of each length. The findings therefore reveal subtle but potentially important adult age differences in the effects of fixation location on Chinese word recognition. Moreover, the similarity in effects for words and nonwords implies a more general age-related change in oculomotor strategy when processing Chinese character-strings.

Visual speech perception in foveal and extrafoveal vision: further implications for divisions in hemispheric projections

When observing a talking face, it has often been argued that visual speech to the left and right of fixation may produce differences in performance due to divided projections to the two cerebral hemispheres. However, while it seems likely that such a division in hemispheric projections exists for areas away from fixation, the nature and existence of a functional division in visual speech perception at the foveal midline remains to be determined. We investigated this issue by presenting visual speech in matched hemiface displays to the left and right of a central fixation point, either exactly abutting the foveal midline or else located away from the midline in extrafoveal vision. The location of displays relative to the foveal midline was controlled precisely using an automated, gaze-contingent eye-tracking procedure. Visual speech perception showed a clear right hemifield advantage when presented in extrafoveal locations but no hemifield advantage (left or right) when presented abutting the foveal midline. Thus, while visual speech observed in extrafoveal vision appears to benefit from unilateral projections to left-hemisphere processes, no evidence was obtained to indicate that a functional division exists when visual speech is observed around the point of fixation. Implications of these findings for understanding visual speech perception and the nature of functional divisions in hemispheric projection are discussed.

An ERP assessment of hemispheric projections in foveal and extrafoveal word recognition

Background

The existence and function of unilateral hemispheric projections within foveal vision may substantially affect foveal word recognition. The purpose of this research was to reveal these projections and determine their functionality.

Methodology

Single words (and pseudowords) were presented to the left or right of fixation, entirely within either foveal or extrafoveal vision. To maximize the likelihood of unilateral projections for foveal displays, stimuli in foveal vision were presented away from the midline. The processing of stimuli in each location was assessed by combining behavioural measures (reaction times, accuracy) with on-line monitoring of hemispheric activity using event-related potentials recorded over each hemisphere, and carefully-controlled presentation procedures using an eye-tracker linked to a fixation-contingent display.

Principal Findings

Event-related potentials 100–150 ms and 150–200 ms after stimulus onset indicated that stimuli in extrafoveal and foveal locations were projected unilaterally to the hemisphere contralateral to the presentation hemifield with no concurrent projection to the ipsilateral hemisphere. These effects were similar for words and pseudowords, suggesting this early division occurred before word recognition. Indeed, event-related potentials revealed differences between words and pseudowords 300–350 ms after stimulus onset, for foveal and extrafoveal locations, indicating that word recognition had now occurred. However, these later event-related potentials also revealed that the hemispheric division observed previously was no longer present for foveal locations but remained for extrafoveal locations. These findings closely matched the behavioural finding that foveal locations produced similar performance each side of fixation but extrafoveal locations produced left-right asymmetries.

Conclusions

These findings indicate that an initial division in unilateral hemispheric projections occurs in foveal vision away from the midline but is not apparent, or functional, when foveal word recognition actually occurs. In contrast, the division in unilateral hemispheric projections that occurs in extrafoveal locations is still apparent, and is functional, when extrafoveal word recognition takes place.

Functional foveal splitting: evidence from neuropsychological and multimodal MRI investigations in a Chinese patient with a splenium lesion

It remains controversial and hotly debated whether foveal information is double-projected to both hemispheres or split at the midline between the two hemispheres. We investigated this issue in a unique patient with lesions in the splenium of the corpus callosum and the left medial occipitotemporal region, through a series of neuropsychological tests and multimodal MRI scans. Behavioral experiments showed that (1) the patient had difficulties in reading simple and compound Chinese characters when they were presented in the foveal but left to the fixation, (2) he failed to recognize the left component of compound characters when the compound characters were presented in the central foveal field, (3) his judgments of the gender of centrally presented chimeric faces were exclusively based on the left half-face and he was unaware that the faces were chimeric. Functional MRI data showed that Chinese characters, only when presented in the right foveal field but not in the left foveal field, activated a region in the left occipitotemporal sulcus in the mid-fusiform, which is recognized as visual word form area. Together with existing evidence in the literature, results of the current study suggest that the representation of foveal stimuli is functionally split at object processing levels.

Evaluating hemispheric divisions in processing fixated words: the evidence from Arabic

Some studies have claimed that, when fixating a word, a precise split in foveal processing produces substantial effects on word recognition because all letters to the left and right of fixation project to different, contralateral hemispheres. Recently in this Journal, Jordan, Paterson, Kurtev, and Xu (2010, Cortex, 46, 298-309) evaluated this claim using precisely-controlled procedures of fixation and stimulus presentation and found no evidence of split-foveal processing. However, in line with other research in this area, these findings were obtained using a Latinate alphabetic language (in this case English) which may induce specific effects on performance. Consequently, here we report a further study which used stimuli from a fundamentally different, non-Latinate alphabetic language (Arabic) with characteristics better suited to revealing effects of split-foveal processing, if they exist. Participants made lexical decisions to five-letter Arabic words (and nonwords) when fixations were made immediately to the right (location 1) or left (location 6) of each stimulus, or at one of the four possible inter-letter locations (locations 2-5). Fixation location was carefully controlled using an eye-tracker linked to a fixation-contingent display and all stimuli were presented within foveal vision to avoid confounding influences of extrafoveal projections. Performance was equally poorest when fixating locations 1 and 6 (i.e., when words were shown entirely to either the left or right of fixation), equally intermediate for locations 2 and 5, and equally superior for locations 3 and 4 (i.e., the centre of words). Moreover, additional, word-specific analyses showed no evidence of the effects of fixation location on individual word recognition also predicted by split-foveal processing. These findings from a non-Latinate language complement those reported previously for English to provide further evidence that while fixation location influences word recognition, these influences occur with no functional division in hemispheric processing at the point of fixation.

Do fixation cues ensure fixation accuracy in split-fovea studies of word recognition?

Many studies have claimed that hemispheric processing is split precisely at the foveal midline and so place great emphasis on the precise location at which words are fixated. These claims are based on experiments in which a variety of fixation procedures were used to ensure fixation accuracy but the effectiveness of these procedures is unclear. We investigated this issue using procedures matched to the original studies and an eye-tracker to monitor the locations actually fixated. Four common types of fixation cues were used: cross, two vertical gapped lines, two vertical gapped lines plus a secondary task in which a digit was presented at the designated fixation point, and a dot. Accurate fixations occurred on <35% of trials for all fixation conditions. Moreover, despite the usefulness often attributed to a secondary task, no increase in fixation accuracy was produced in this condition. The indications are that split-fovea theory should not assume that fixation of specified locations occurs in experiments without appropriate eye-tracking control or, indeed, that consistent fixation of specified locations is plausible under normal conditions of word recognition.