a, Bimodal distribution of spike width of extracellularly recorded units (n = 558 from 4 animals, see Methods for details). Spike width less than 0.775 ms (dash line) was used as a criterion for fast spiking interneurons. b, Mean spiking rate inversely correlated with spike width. Putative fast spiking interneurons (red) were identified as spike width<0.775 ms and mean spiking rate > 1 Hz. Inset: averaged spike waveform of putative fast spiking interneurons (red) and pyramidal cells (black). Note that spikes from putative fast spiking interneurons decayed faster. SD was plotted as shading areas. Scale bar: 0.5 ms. The vast majority of recorded units (n = 455/558 units) show spike waveforms that exhibit larger negative peaks than the positive one. Only these units were used to calculate the inset figures to avoid distortion of spike waveforms when both polarities were averaged together. c, Spatial rate map of putative fast spiking interneurons. Cells were ranked based on the COM of their firing fields. d, Averaged population activity of putative fast spiking interneurons on oval (black) and triangular (red) tracks. e, Mean spiking rates on oval and triangular tracks (within the shared cue region, between grey lines in e) are not significantly different in fast spiking interneurons (p = 0.471, n = 66 cells from 4 animals). f–i, The same as (b–e) with more strictly defined putative fast spiking interneurons (units distributed at the upper-left corner in f). Mean spiking rates of this population were not dependent on context either (p = 0.8939, n = 17 cells from 4 animals). Solid curves and shaded areas depict mean±s.e.m. Bar graphs and error bars represent mean and s.e.m., respectively. Paired student t-test (two-sided) was conducted in all statistical analyses. N.S.: not significant (P > = 0.05).