Event-related gamma-band activity in visual object representation

Visual object representation is a prime example of the human brain’s ability to efficiently and rapidly process highly complex and ambiguous information. Object processing has always been one of the central topics of cognitive neuroscience, but the swift and interactive neural mechanisms through which it is achieved are still not well understood. High frequency oscillatory activity, registered in the gamma-band of the human EEG (30-80Hz), plays an important role in the perceptual organisation of a wide range of visual properties into coherent scenes. The aim of this study was to demonstrate how such activity could be a possible mechanism subserving a range of processes crucial for visual object representation. Therefore, evoked and induced (i.e., not time- or phase-locked) gamma-band activity (GBA) were researched in a set of experiments that examined the mnemonic, perceptual and attentional side of visual object representation. Building up from the representational hypothesis approach, it was assumed that evoked GBA would reflect early perceptual processing of image-features, occurring up to 100ms post-stimulus onset, while induced GBA would reflect a late mnemonic continuation of representational activity lasting up to approx. 300ms. In an initial set of experiments, aimed at the mnemonic aspects of object recognition, it was shown that induced GBA peak latency delays are predictive of recognition delays in view-dependent object processing. The following set of experiments focused on perceptual representational processing by examining the effects of different object image-properties. It was found that while evoked GBA responded by monotonous amplitude increases related to the quantity of features in the image, induced GBA peak latency was sensitive to the quality of the features, being delayed in conditions where extra detail increased the difficulty of image decoding. The last experiment focused on the automaticity of object processing reflected in the event-related GBA, approaching it from the perceptual load framework. Low and high load tasks where co-presented with task-irrelevant familiar or unfamiliar objects. It was found that evoked GBA amplitude was mostly responsive to task complexity, while induced GBA showed object-specific amplitude enhancements under low load. We conclude that event-related GBA in visual object representation subserves dynamic object coding processes: evoked GBA is related to low-level integration of features and their conjunctions, while induced GBA reflects a high-level mnemonic continuation of representational activity.