The temporal context of face perception
A given picture of a person might look different at different times. Previous encounters with other people or in other words, the temporal context of a given face modifies its perception. Our previous research on priming and adaptation is shaping theories about the neural mechanisms and representations involved in face perception. As we are beginning to understand the relationship between these two phenomena, an account of the interaction between top-down processes, such as predictions, attentional cueing and sensory competition among stimuli becomes increasingly important. In this project proposal, we will continue to study the effect of previous experiences on face perception using psychophysical, electrophysiological and neuroimaging methods within the theoretical frame of predictive coding models.  
In the first line we capitalize upon the previously found interactions among multiple simultaneously presented faces (Nagy et al., 2011). Using event-related potential (ERP) recordings we test the temporal dynamics of sensory competition among faces. We compare the competition effect (manifested in the reduction of blood oxygen level dependent (BOLD) signal) for different stimulus categories compared to faces and test if a prior stimulus, serving as an attentional cue is able to bias these competitions similarly for various categories or not. We test competition effects among faces and body-parts and use functional magnetic resonance imaging (fMRI) to evaluate how their neural coding changes during the perception of an entire person as a “Gestalt”.  

In the second line of (mostly fMRI) experiments we concentrate on the interaction of predictive cueing, attention and adaptation. We try to disentangle the effect of expectation from that of passive stimulus probability by manipulating both effects orthogonally. Second, using ambiguous stimuli (Cziraki et al., 2010) we test whether prediction affects the neuronal responses per se or rather only their suppression, due to adaptation. Finally, we test how predictive cueing affects sensory competition among faces and other stimulus categories.
Prof. Dr. Gyula Kovács
Géza Gergely Ambrus, PhD
Charlotta Eick, MSc
Sophie-Marie Rostalski, MSc
Kathrin Wiese
Predictive processes in the brain
The major objective of the present grant proposal is to better understand how the brain predicts future events by testing a multi-stage model of predictive coding (PC). PC is one of the theories of evoked brain responses which postulates that vision is a hierarchical process in which higher order areas shape and predict the tuning properties of lower level areas via strong feedback connections. This is achieved by suppressing the predicted, and hence redundant, neural responses in lower level areas that are consistent with the higher-level expectations, resulting in a suppressed neural response and in an efficient encoding mechanism. 
For understanding perceptual predictions better we will compare the basic paradigms of predictive studies, using functional magnetic resonance imaging techniques. We will apply variations of paradigms modulating stimulus probabilities (including passive oddball paradigms) and explicit cue-stimulus/stimulus-stimulus associations that lead to predictions of future events. We also will apply a newly developed paradigm that manipulates implicit stimulus probabilities and explicit expectations orthogonally to test the different levels of the PC model. As part of this experimental series we will test if surprise related augmentation; repetition related suppression or both are responsible for the previously observed predictive effects. 
We also will apply an extensive perceptual learning task with non-face stimulus categories and test if this training can change the modulatory effects of predictions. In addition, we will test if passive exposure in typical statistical learning paradigms is satisfactory or an active task and perceptual learning is required for observing predictive effects. We also will recruit perceptual expert subjects, such as car experts and test if/how long-term and extensive experience with such stimulus categories alters the predictive modulations of the neural responses. Overall, our aim is to reveal the neural mechanisms of future event predictions in the human brain and give evidences of the multiple-level hierarchical model of predictive functions.
The hierarchy and parallelity of the visual face processing network
Human face - a stimulus that is the continuous source of information about individuals around us. Gender, age, emotional state, trustworthiness, beauty are just a few examples of the many important properties of a face which determine our social life. Thus it is not surprising that a great amount of experimental, social psychological, cognitive and systemic neuroscientific experiments are performed every year to describe the details of face processing in the human brain. One of the (if not the single) most important task of a face is to convey information about the identity of others or in other words to support recognition. We recognize the faces of our relatives, colleagues, neighbors, shopkeepers, our favorite Hollywood actors/actresses. Despite its obvious importance and the large amount of studies published yearly, the neural background of face recognition are still unclear as of today. 
The major aim of the current proposal is to test, by using psychophysical methods together with transcranial magnetic stimulation (TMS) and functional magnetic resonance imaging (fMRI) methods how the occipito-temporal and anterior-temporal cortical areas of the human brain enable us to acquire and maintain a stable representation about known, familiar persons, thereby supporting face recognition. 
To appreciate the importance of the topic and the novelty of the proposed approach, first I briefly summarize the currently available theories of face processing, with special attention to face recognition. I also describe briefly the central hypothesis of the proposal: instead of having one single face recognition unit or area, a hierarchical, but at the same time distributed, recurrent network is responsible for face recognition. Finally, I describe the differences between familiar and unfamiliar face representations and summarize the available data on the transition process from unfamiliar to familiar representations.