CRS Guest Talks, Best Poster Prizes & Travel Awards > CRS Guest Lecturers > Jasna Martinovic
Jasna Martinovic is a Senior Lecturer in Psychology at the University of Aberdeen.
How does basic visual information that we sample from the environment in the form of luminance and colour contrast get transformed by our visual system into a representation of our environment? This might seem a trivial question, but the rich world of objects that we experience in everyday life is derived from basic signals about brightness and chromaticity which subsequently get processed by structurally and functionally complex areas of our brain.
My work concerns object representation, in particular its relation to the synergistic processing of luminance and chromatic signals. I am interested in how this joint processing influences perceptual organisation of scenes, attentional selection and object representation. I use a combination of psychophysical and EEG methods in my work. Psychophysics allows me to measure levels of contrast necessary to perform various visual tasks, or levels of other stimulus properties such as depth or coherence that lead to certain percepts being formed. On the other hand, EEG provides a window into rapidly occuring neural processes that relate to these visual percepts.
Conferenza del Colore 2014: Interactions between Colour and Luminance Signals in Object Classification
The visual system processes objects embedded in complex scenes that spatially vary in both luminance and colour. However, most models of object recognition posit that the processing of object shape is mainly driven by information provided by luminance-driven spatial frequency channels. To determine if chromatic signals also contribute to form processing, the interdependence of activity within the luminance (L + M) and opponent chromatic (L − M and S − [L + M]) postreceptoral mechanisms was investigated in mid-level and high-level vision. Mid-level processes extract contours and perform figure-background organization whereas high-level processes depend on additional semantic input, such as object knowledge. While the S - (L + M) mechanism was not expected to contribute to these spatial vision processes due to its low resolution, L-M signals were expected to combine with co-localised luminance signals in a facilitatory fashion. In the first series of experiments, mid-level and high-level discrimination contrast threshold data was collected over a range of conditions that isolate mechanisms or simultaneously stimulate them. Contrast-dependent, facilitatory interactions between the luminance and L − M inputs were found, while S − (L + M) signals did not interact with luminance. In the second line of experiments, high-level thresholds were remeasured in the presence of backgrounds. Thresholds were elevated when a background was present only for combined S-(L+M) and L+M signals. This suppressive effect increased for a denser background and was dependent on the presence of relatively high S-(L+M) contrast. The same effect was revealed using a crowding paradigm, indicating that a possible mechanism for the suppressive effect of S - (L + M) signals is due to S-cones' larger receptive fields promoting texturisation by offering a larger integration window. In conclusion, colour signals can and do play an important role in the processing of object form. Therefore, current models of object recognition should be revised in order to also incorporate signals from chromatic channels at different spatial frequencies.