• A
  • A
  • A
  • ABC
  • ABC
  • ABC
  • А
  • А
  • А
  • А
  • А
Regular version of the site

Interdisciplinary Studies of Vision

Priority areas of development: humanitarian

Computational modeling is an important and effective field of research in psychology, that allows us to instantiate and test many popular theories about the brain and cognitive functions. If we consider the brain as an information processing system, then there must be a way by which it select and prioritize information when it exceeds the capacity of that system. This mechanism of selection, inhibition and prioritizing of incoming information is called ‘attention’. With the introduction of computational models as a methodology within cognitive sciences, we are able to combine the results of multiple studies about the brain and cognitive functions, while still being able to predict specific results and generate new hypotheses to be tested.

The purpose of this project is to create and test the first version of an integrated computer model that simulates the operation of the human visual and oculomotor systems. Our goal is to develop a model that simulates the flow of information processing from the “input” of visual information in the form of a retinal image to the “output” in the form of gaze fixations and saccadic eye movements.  The resulting model should maintain biological plausibility of its component parts as well as model known cognitive mechanisms such as early facilitation, inhibition of return and anti-saccadic eye movements.   We believe that combining these goals produce the best of results within computational cognitive neuroscience.

To this end we conducted multiple new experiments using techniques ranging from behavioural, eye tracking, TMS and MEG.  Experimental results are important to understand our key attentional processes, but also to generate data useful for our model.  Four new pilot experiments include:

  • using the search task (number of participants n = 15),
  • research in the paradigm “Space invaders” (n = 15),
  • visual re-mapping (n = 5),
  • motion detection experiment (n = 7).

The first three studies were conducted using eyetracking, and the last included trans cranial magnetic stimulation (TMS). All experiments involved healthy adult subjects with normal or corrected vision.

In addition, we completed data collection for 5 studies with a finalized protocol:

  • study of the latent shift of attention (number of participants n = 20),
  • search task (n = 40),
  • visual re-mapping (n = 20),
  • anti-saccade task (n = 20 Russia + n = 20 Iceland)
  • the task of choosing from two alternatives (n = 20).

For modelling results, we implemented code for 7 different classifiers, two models of early processing of visual information in the visual cortex, two models of abstract (non-spatial) generation of saccades and two models of saccadic generation in two-dimensional space. Two saccade generation algorithms were implementations of our own theory:  a temporal diffusion model and a spatial leaky competing accumulator model.

In general, our results for this year reflect great progress in creating a complete simulation of the processing of visual information. We chose the exact implementation of the spatial component, identified the main shortcomings of most currently existing time models and obtained the first results on our two solutions to the problem of modeling temporal dynamics. We created a full processing stream model and used this model to test three popular conflicting theories of attentional capture. We also successfully tested the predictive power of the model in relation to classical phenomena of attention, such as inhibition of return and programming of anti-saccades, on the data obtained during experimental studies of the Laboratory.


Krasovskaya S., Zhulikov G., MacInnes W. Deep Learning Neural Networks as a Model of Saccadic Generation // Российский журнал когнитивной науки. 2019. P. 1-10. 
Krasovskaya S., Кристьянссон А., MacInnes W. Poster Presentations, THE EFFECT OF THE ANTISACCADE TASK ON MICROSACCADE SUPRESSION IN THE POSNER CUEING PARADIGM, in: Book of abstracts: XVI European congress of psychology (ECP 2019) (2‒5 July, 2019, Lomonosov Moscow State University, Moscow). Moscow : Издательство Московского университета, 2019. С. 1866-1866. 
Martinovic J., Paramei G., MacInnes W. Russian blues reveal the limits of language influencing colour discrimination // Cognition. 2020. Vol. 201. P. 104281-. doi
Merzon L., Malevich T., Zhulikov G., Krasovskaya S., MacInnes W. Temporal limitations of the standard Leaky integrate and fire model // Brain Sciences. 2020. Vol. 10. No. 1. P. 1-19. doi
Krasovskaya S., Zhulikov G., MacInnes W. Deep Learning Neural Networks as a Model of Saccadic Generation / Center for Open Science. Series PsyArXiv Preprints. "нет". 2021.