Fig. 5: Omission responses are not explained by stimulus-driven response dynamics.
From: The representation of omitted sounds in the mouse auditory cortex
A We compared neural responses to omissions in a standard sequence with an SOA of 200 ms to non-omissions in a sequence with an SOA of 400 ms. The identical pause between stimuli is the normal, expected interval in SOA 400, but is an unexpected deviation in SOA 200, and therefore controls for basic stimulus-driven response dynamics, such as offset responses. B SOA 200 omissions responses (OR, light gray) were substantially larger than SOA 400 interstimulus responses (ISR, dark gray). C The difference between OR 200 and ISR 400 localized to a very similar area as the OR 200 itself. Across trials (D) in the animal shown in (C), as well as across animals (E) and cells from the ORR of all recorded animals (F), the same behavior was observed. Significance: p < 0.05 (*), p < 0.01 (**), p < 0.001 (***) (see Supplementary Data 1 for statistical details). G We checked whether a second omission in the SOA 200 sequence led to a further increase in the neural response by introducing double omissions into the sequence of standard tones. H Comparing the primary region (PR) with the omission-responsive region, a timed increase in neural activity occurs at the onset of the second omission on top of the elevated response from the first omission for a representative animal. Mean with ±SEM shading. Same for (M and N). I An index of the additional increase to the second omission localized as well to the omission area, i.e., the medial and posterior parts of TeA. J Cluster analysis of the second omission response confirms a location consistent with the ORR, while largely absent in the PR. K The double omission paradigm was run in 5 animals, each of which showed a further increase in response to the second omission relative to the first, normalized to the second omission response size for clarity. L We presented multiple SOAs, ranging from 100–400 ms, to investigate the dependence of the response dynamics on the time of expected occurrence of the next stimulus. M, N The OR showed different time-courses depending on the SOA, faster for short SOAs, slower for longer SOAs, in single animals, as well as the average across animals (N, n = 5). O The OR showed a bandpass-like dependence on SOA, similar to ORs in previous studies in humans, rats, and mice20,25,26. P The average slope of the OR significantly decreased with longer SOAs (p = 0.0011, Kruskal–Wallis test). Q The minimum value of SR shows a significant decrease from SOA = 100 ms to longer SOAs (p = 0.001, Wilcoxon rank sum test). Source data are provided with this paper.
