Abstract
Response properties of regions in ventral temporal cortex (VTC) have often been characterized during highly controlled experimental paradigms where an unpredictable sequence of single, isolated objects are presented for a few hundred milliseconds to passively fixating participants. The visual environment of the real world is relatively stable and predictable, containing contextual information that grounds objects in their environment, which is actively and volitionally sampled during free-viewing. The degree to which these differences between experimental versus naturalistic viewing conditions are reflected in the neurodynamics of VTC regions remains unclear. To compare these responses, we presented natural scenes to one pilot patient with intractable epilepsy during a free-viewing paradigm while collecting electrocorticography (ECoG) from VTC and eye tracking data. The VTC neurodynamics collected during free-viewing of natural scenes was compared to the responses evoked by similar objects presented in a more traditional experimental paradigm. VTC data could be used to reliably predict when the subject was looking at a face or not during free-viewing. Furthermore, above-chance decoding occurred as early as 200 ms prior to saccades landing on a face. This suggests that face-processing occurs in VTC even prior to fixating faces in natural scenes. Classifiers trained to discriminate face from non-face trials in the traditional paradigm could reliably discriminate face from non-face fixations in the free-viewing task. This suggests that neural responses during the traditional experimental paradigm shared some similarity to those evoked during free-viewing paradigm. However, responses evoked by face fixations in the free-viewing task were stronger in the high gamma frequencies (40-200 Hz) and less pronounced in lower frequency ranges compared to those evoked by the traditional paradigm. Future work is necessary to investigate how image characteristics, such as scene context, contribute to pre-saccadic activation of object-selective VTC and whether these results generalize to larger patient populations.