Schematic of the paradigm used to measure integration windows. Segments of natural stimuli are presented using two different random orderings (concatenated using cross-fading). As a consequence, the same segment is surrounded by different context segments. If the segment duration is longer than the integration window (top panel), there will be a moment when the window is fully contained within each segment. As a consequence, the response at that moment will be unaffected by the surrounding context segments. If the segment duration is shorter than the integration window (bottom panel), the integration window will always overlap the surrounding context segments, and they can therefore alter the response. The TCI paradigm estimates the minimum segment duration needed to achieve a context invariant response. This figure plots waveforms for an example sequence of segments that share the same central segment. Segment boundaries are demarcated by colored boxes. The hypothesized integration window is plotted above each sequence at the moment when it best overlaps the shared segment.