News & Views (Y, AA, BB)

643 In everyday life, we operate on a very broad range of time scales, from sub-milliseconds to years. For instance, we perceive two clicks that are closer together than 50 milliseconds as a single event, even though we detect the timing differences. (We use them to localize sounds in space and to discriminate between syllables in speech.) If the same clicks are presented repetitively at intervals from 100 milliseconds to 2 seconds, we perceive them as rhythms. Presented at intervals longer than 2 seconds, the same clicks seem independent, with no discernible rhythm1. Psychophysical studies suggest that we estimate these longer intervals— seconds to hours—either by estimating intervals or by referring to our internal circadian clock2. These time estimates tend to be similarly accurate whether they are tested by cognitive responses (“it has been three minutes now”) or by making movements, such as tapping our fingers at regular intervals3. Surprisingly, neuroscientists have been much more successful at explaining the hows and whys of timing at the sub-millisecond intervals than in finding out how nervous systems judge and produce the 0.5to 2-second intervals that we overtly recognize as rhythmic events. The paper by Scott Hooper4 tackles this issue headon, suggesting a possible neuronal mechanism for determining and producing rhythms. The mechanism proposed is based on his physiological studies of the properties of individual neurons in the lobster stomatogastric ganglion; these neurons normally help to produce rhythmic motor patterns that control movements of the animal’s gut. Their firing is influenced by (hence, ‘encodes’) both the timing of repeating pulses and their duration. He proposes a way to decode this through NM, but axons entering NL have very precise and orderly conduction differences: they vary slightly in diameter, and therefore in conduction time. These are the ‘delay lines’. This means that certain NL neurons receive input from the left ear (where the sound arrived later) via axons that are larger than those from the right ear, so the signals from the two ears arrive at exactly the same time at one NL neuron, such as E; simultaneous input is necessary for activation of those neurons (termed ‘coincidence detectors’). The conduction delay differences required to make this coincidence-detection scheme work is remarkably short: on the order of 0.01 to 0.1 milliseconds! Our perception of speech depends upon measuring timing differences at He’s got rhythm: single neurons signal timing on a scale of seconds