When universal visuo-motor mechanisms meet endogenous and language-related processes: Determining principles of eye-movement control in reading. – UniversalRead2

The first goal was to determine the visuo-motor mechanisms that drive the eyes along the lines of text during reading. To this aim, three series of experiments were conducted. The first, parametric set of studies characterized the global effect obtained when a peripheral visual target (i.e. a simple shape) is displayed with or without a distractor stimulus. The second set of studies tested the effect of visual factors on the distributions of the eyes' initial landing positions in isolated words. The third, corpus-type study aimed at directly comparing the characteristics of saccades in simple saccade-target tasks and natural reading on the same set of participants. This third study is not completed yet. For the second, modeling part of the project, the German team revised their Bayesian model in order to account for variations of the eyes' initial landing positions in words with the spatial arrangement of prior letters. The French team used an existing model of the Superior Colliculus in monkeys to derive the neuronal activity profile that results from different types of visual stimulations (simple shapes and words). The third goal of the project was to study the representations of the meaning of words and the respective roles of semantics and syntax in reading. To this aim, several behavioral studies using multiple semantic priming were conducted. A cortical model of semantic priming was then further developed to account for the obtained data. In addition, a pilot study was conducted in order to distinguish the role of semantic and sentential predictibility. The experiment is now ready and will be run before the end of the project.

In relation with the first goal of the project, we showed that (1) the central foveal zone is a deadzone for the global effect (Casteau & Vitu, 2012), (2) the maximal distance betweeen two stimuli for a global effect increases with stimulus eccentricity, but is relatively invariant if the distortion of visual space in oculomotor-center maps is taken into account (Casteau & Vitu, 2011-COM; en preparation), (3) lateral inhibitions are at work in oculomotor-center maps (Tandonnet, Massendari & Casteau, 2012), (4) the global effect depends more greatly on the visual salience of the stimuli (their size or number of elementary features) than on the location of their boundaries (Massendari, Tandonnet & Vitu, 2013-COM; en préparation; Tandonnet, Casteau & Vitu, 2013; Tandonnet & Vitu, 2013), (5) character size and the spatial arrangement of the letters preceding a word influences the distributions of the eyes' initial landing positions in words (Krügel, Vitu & Engbert, 2012; Yao-N'Dré, Castet & Vitu, soumis). In relation with the second goal of the project, the German team showed that a revised version of their Bayesian model can account for variations of the distributions of the eyes' initial landing position in words with the letter spatial arrangement. The French team showed that the systematic undershoot of eccentric targets is simply the result of the distortion of visual space in oculomotor-center maps (Vitu & Casteau, 2013-COM; en préparation). In relation with the 3rd goal of the project, the conducted studies showed that (1) the semantic-priming shift, the fact that the recognition of a word tends to be facilitated by one or the other previously read words depends on the strength of association between each prime and the target (Lavigne, Dumercy, Chanquoy, Mercier & Vitu-Thibault, 2012), and (2) the time course of the semantic priming shift depends not only on association strength, but also on reading time per word (Lavigne, Chanquoy, Dumercy & Vitu, 2013).

First, the corpus data collection and analysis will be completed in order to compare the properties of saccades during the execution of simple saccade-target tasks and natural reading. This part of the project is time consuming, but very promising (and also useful for Goal 3 of the project).
Second, the data showing that the systematic tendency to undershoot far targets can be explained by the distortion of visual space in oculomotor-center maps will be published. These data will then form the basis of a neural model of eye movements in reading, that will be developed even after the end of the contract.
Third, data will be collected and analyzed in order to distinguish the role of semantics and syntax during sentence reading.

1. Casteau, S. & Vitu, F. (2012). On the effect of remote and proximal distractors on saccadic behavior: A challenge to neural-field models. Journal of Vision, 12(12):14, 1-33.
2. Krügel, A., Vitu, F., & Engbert, R. (2012). Fixation positions after skipping saccades: A single space makes a large difference. Attention, Perception & Psychophysics, 74(8), 1556-1561. DOI 10.3758/s13414-012-0365-1.
3. Lavigne, F., Chanquoy, L., Dumercy, L. & Vitu, F. (2013). Early dynamics of the semantic priming shift. Advances in Cognitive Psychology, 9(1), 1-14.
4. Lavigne, F., Dumercy, L., Chanquoy, L., Mercier, B. & Vitu-Thibault, F. (2012). Dynamics of the semantic priming shift: Behavioral experiments and cortical network model. Cognitive Neurodynamics, 6, 467-483. DOI 10.1007/s11571-012-9206-0
5. Lavigne, F., Dumercy, L. & Darmon, N. (2011). Determinants of Multiple Semantic Priming: A Meta-Analysis and Spike Frequency Adaptive Model of a Cortical Network. The Journal of Cognitive Neuroscience, Vol. 23, No. 6: 1447–1474.
6. Tandonnet, C., Casteau, S., & Vitu, F. (2013). On the limited effect of stimulus boundaries on saccade metrics. Journal of Vision.
7. Tandonnet, C., Massendari, D. & Vitu, F. (2012). When larger visual distractors become less disruptive: Behavioral evidence for lateral inhibition in saccade generation. Journal of Vision, 12(4):2, 1–11, www.journalofvision.org/content/12/4/2, doi:10.1167/12.4.2.
8. Tandonnet, C. & Vitu, F. (2013). Stimulus properties and saccade metrics: When local features are more critical than global features. Behavioral Neuroscience, 127(1), 121-125.

Reading is probably the most complex cultural skill which is made possible by the successive movements of our eyes along the lines of text. The present research proposal is the result of a first ANR-DFG project (2007-2010) and hence three years of joint research effort between the French and the German partners on the psychology of reading. We contributed significant new results to the fields of reading, eye-movement control and psycholinguistics and demonstrated how psychological research on the reading process may benefit from computational modelling and the neurosciences of active vision and linguistics. These new interdisciplinary and integrated research questions will be addressed here, benefiting from the different and complementary expertises of the researchers involved. The first goal is to specify the basic universal visuo-motor mechanisms that underlie reading in order to elaborate a neurophysiologically plausible model based on experimentation (from simple oculomotor tasks to word reading) as well as mathematical modelling. We will revise and extend our initial Center-of-Gravity (CoG) model of initial landing sites within words. The second goal is to determine experimentally and computationally how the CoG model can be reconciled with evidence for Bayesian estimation (discovered in the first project) as well as attention-based eye guidance. The third goal is to determine how Human-specific language processes combine with universal visuo-motor constraints, expectations, and attention, with particular emphasis on the interaction between semantics and syntax. In perspective, this multi-level approach of reading might help setting up the bases for early dissociation of oculomotor, expectation-based, attentional, and language-related components of reading difficulties and dylexia.