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

About the Laboratory

Computational modelling has brought a new way to instantiate and test many popular theories of brain and cognition. The cognitive revolution in psychology and neuroscience opened a new paradigm using experimental and neuroimaging methods. Although our understanding at the micro-experimental level improved, it was difficult to incorporate these results into larger macro-theories. Computational modelling allows us to integrate results while still maintaining the experimental precision of individual results and to make new testable predictions. Many cognitive models now match human data on behavioural tasks and neural models can match imaging data. However, we believe this should only be the first step.

Deep Learning Neural Nets (DLNNs) have been extremely successful in performing automated vision tasks such as still image classification. DLNNs, as their name suggests, contain many layers of similarly structured hidden nodes, with each layer trained to detect increasingly complex features within images. These layers, at least visually, can be seen as mirroring the way the human visual cortex extracts simple features starting at V1 and progressively more complex structure, objects and spatial maps. We know that these models perform visual classification tasks as well humans (and sometimes better). For example, DLNNs have recently taken the top scores on the MIT saliency benchmark. What hasn’t been shown yet, is whether these algorithms are good models of the human visual system in how they process visual information.

Computational models have potential to transform science and medicine, by producing simulations that respond to intervention in ways similar to human participants. Although these simulations may never be perfect, they have the possibility to predict which interventions (drugs, treatments, tasks, interfaces, etc.) will produce the desired effect and minimize the time and resources needed by focusing on the most likely interventions.

The Vision Modelling Laboratory is actively engaged in creating highly predictive and biologically plausible models for processing visual information and generating saccades. The laboratory is a permanent platform for experimental and computational research of the human visual system, as well as a space for cooperation in the scientific and educational fields of research centres, laboratories and groups operating at the Faculty of Social Sciences of HSE, bringing together researchers engaged in eye tracking, cognitive neuroscience and computational modeling.


 

Have you spotted a typo?
Highlight it, click Ctrl+Enter and send us a message. Thank you for your help!
To be used only for spelling or punctuation mistakes.