Diagram of 2-choice electronic-echo delay discrimination set-up with bat on Y-shaped platform: Bat’s broadcast sounds are picked up by microphones (m), delayed and filtered electronically, and returned as echoes of virtual targets from loudspeakers (s). Delay of is fixed at , which simulates a target range of 54.5 cm. Delay of S− is fixed later, at , which simulates a target range of 68.3 cm. The difference between S− and corresponds to the index point described in Fig. 2. For the present experiments, and S− undergo various filtering conditions to assess effects on performance at index point. Idealized percent-error performance curve illustrates relation between peak in error curve when S− matches delay of and the still-elevated performance occurring at .
Experiment 1. Effects of removing harmonics on delay acuity: (a) Spectrograms for stimuli used as echoes of a representative bat broadcast containing FM1 and FM2. Highpass filtering restricts the echo band to 20–90, 28–90, 32–90, or 66–90 kHz, which removes progressively larger segments from the low-frequency end of FM1. Lowpass filtering to 20–44 kHz completely removes FM2, leaving FM1 mostly intact. Filters have roll-off of 115 dB/octave to sharply cut out the unwanted part of the echo spectrum. Both and S− are filtered as indicated and then delivered with the index delay difference of (Fig. 3). (b) Bar-graph showing performance on delay-discrimination tasks (percent errors) for four bats (150 trials/bat), with mean performance ±1 SD (circles). Removal of FM1 in small stages causes progressive, significant loss of delay acuity (more errors) for difference between and S− , culminating in chance performance for echoes containing only FM2. Performance recovers but still is not completely normal when echoes contain only FM1.
Experiment 2. Relative salience of harmonics: (a) Spectrograms for representative stimuli containing only FM1 or only FM2 for and S−. Full-band, normal echoes are shown at left. At right, contains only FM2 while S− contains FM1 and FM2 (full band). (b) Bar graph showing performance on delay-discrimination tasks (percent errors) for four bats with mean performance ±1 SD. As already shown in Fig. 2, performance on delay difference is good if echoes contain only FM1 but chance if echoes contain only FM2. When offered a choice between S− containing FM1 and FM2 versus containing FM2 alone, bats prefer S− even though it is delivered later than .
Experiment 3. Effects of harmonic-split filtering on delay acuity: (a) Spectrograms for representative stimuli testing the effect of introducing the harmonic-split filters on the index delay discrimination task. Full-band, normal echoes with no harmonic split for and S− are shown at left (note slight overlap of harmonics at 50–55 kHz). Next, to split the harmonics into two parallel electronic channels, echoes for both and S− are filtered at 20–44 kHz to remove FM2 and 66–90 kHz to remove FM1. The isolated harmonics are delayed and amplified differently, then recombined (added) to create split-harmonic echoes that now have a narrow spectral gap eliminating frequencies around 55 kHz where FM1 and FM2 overlap. echoes have either the harmonic split alone (FM1 and FM2 both at ) or the harmonic split plus offset of FM2 relative to FM1 (FM1 at , FM2 at ). S− echoes have the harmonic split alone (FM1 and FM2 at ). (b) Bar graph showing performance on delay-discrimination tasks (percent errors) for four bats (150 trials/bat, with mean performance ±1 SD shown by circles). Introduction of harmonic-split filters with no other change causes small, significant decrease in acuity (more errors) compared to full-band echoes . Addition of offset to FM2 relative to FM1 causes large decline in acuity beyond what occurs for split-harmonic condition with no delay difference .
Experiment 4. Amplitude-latency trading and compensatory time shift: (a) Spectrograms for representative stimuli used to assess the loss in delay acuity if FM2 is decreased by 3 dB relative to FM1, and for testing the role of amplitude-latency trading by advancing FM2 in time by a compensatory . (Amplitude-latency trading ratio for bats is .) The two spectrograms at left show full-band, normal echoes with no harmonic split for and with harmonic split for (from Fig. 4). Next two spectrograms show FM2 decreased in strength by 3 dB relative to FM1, or decreased by 3 dB and moved earlier in time by . All S− are full-band echoes with no harmonic split (at right). (b) Bar graph showing performance on delay-discrimination tasks (percent errors) for four bats (150 trials/bat, with mean performance ±1 SD shown by circles). Introduction of harmonic-split filters with no other change causes small, significant decrease in acuity (more errors) compared to full-band echoes . Decreasing FM2 by 3 dB causes further, large decline in performance , but advancing FM2 in time by compensates for amplitude-latency trading and restores performance . Results show no evidence for perceptual effect from 3 dB loss of FM2 except for the associated latency shift.
Electronic filter settings and attenuation of stimuli for four experiments.
Article metrics loading...
Full text loading...