Figure 2: Behavioural task and speech intelligibility results.

(a) Trials consisted of three steps: BEFORE, MIDDLE and AFTER. In the first step called BEFORE (left column), subjects heard a filtered speech stimulus that lacked the key modulations for speech intelligibility. Stimuli were filtered either with a spectral modulation filter (top), removing spectral envelope modulations above 0.5 cycles/kHz or a temporal modulation filter (bottom), removing temporal envelope modulations above 3 Hz (see ‘Methods’ section). In the second step called MIDDLE (centre column), subjects heard the unfiltered version of the spoken sentence. A subset of three subjects had a 50% chance of hearing either the unfiltered version or pink noise with a matched frequency power spectrum. In the third step called AFTER (right column), the same filtered speech stimulus was repeated. Subjects attended to a fixation cross presented during each stimulus and passively listened to the presented sounds. (b) In a separate behavioural task, non-clinical subjects were asked to type any words they heard after the first filtered speech presentation (BEFORE and here labelled no context), after a filtered speech sentence that followed a different unfiltered sentence (AFTER with wrong context), or after a filtered speech sentence that followed the matching unfiltered sentence (AFTER with right context). Mean±s.e. % words correct is shown. More details and results obtained using other contextual stimuli to further explore the stimulus information required for the perceptual enhancement can be found in Supplementary Fig. 1 and ‘Methods’ section.