A comprehensive computational model of animal biosonar signal processing

Echo detection thresholds in big brown bats (Eptesicus fuscus) vary with echo spectral content

Echolocating big brown bats (Eptesicus fuscus) broadcast downward frequency-modulated sweeps covering the ultrasonic range from 100-23 kHz in two harmonics. They perceive target range from the time delay between each broadcast and its returning echo. Previous experiments indicated that the bat's discrimination acuity for broadcast-echo delay declines when the lowest frequencies (23-35 kHz) in the first harmonic of an echo are removed. This experiment examined whether echo detection is similarly impaired. Results show that detection thresholds for echoes missing these lowest frequencies are raised. Increased thresholds for echoes differing in spectra facilitates the bat's ability to discriminate against clutter.

Temporal coherence of harmonic frequencies affects echo detection in the big brown bat, Eptesicus fuscus

Echolocating big brown bats (Eptesicus fuscus) broadcast frequency modulated (FM) ultrasonic pulses containing two prominent harmonic sweeps (FM1, FM2). Both harmonics typically return as echoes at the same absolute time delay following the broadcast, making them coherent. Electronically splitting FM1 and FM2 allows their time delays to be controlled separately, making them non-coherent. Earlier work shows that big brown bats discriminate coherent from split harmonic, non-coherent echoes and that disruptions of harmonic coherence produce blurry acoustic images. A psychophysical experiment on two trained big brown bats tested the hypothesis that detection thresholds for split harmonic, non-coherent echoes are higher than those for coherent echoes. Thresholds of the two bats for detecting 1-glint echoes with coherent harmonics were around 35 and 36 dB sound pressure level, respectively, while thresholds for split harmonic echoes were about 10 dB higher. When the delay of FM2 in split harmonic echoes is shortened by 75 μs to offset neural amplitude-latency trading and restore coherence in the auditory representation, thresholds decreased back down to those estimated for coherent echoes. These results show that echo detection is affected by loss of harmonic coherence, consistent with the proposed broader role of coherence across frequencies for auditory perception.

Oscillatory discharges in the auditory midbrain of the big brown bat contribute to coding of echo delay

Subsequent to his breakthrough discovery of delay-tuned neurons in the bat's auditory midbrain and cortex, Albert Feng proposed that neural computations for echo delay involve intrinsic oscillatory discharges generated in the inferior colliculus (IC). To explore further the presence of these neural oscillations, we recorded multiple unit activity with a novel annular low impedance electrode from the IC of anesthetized big brown bats and Seba's short-tailed fruit bats. In both species, responses to tones, noise bursts, and FM sweeps contain long latency components, extending up to 60 ms post-stimulus onset, organized in periodic, oscillatory-like patterns at frequencies of 360-740 Hz. Latencies of this oscillatory activity resemble the wide distributions of single neuron response latencies in the IC. In big brown bats, oscillations lasting up to 30 ms after pulse onset emerge in response to single FM pulse-echo pairs, at particular pulse-echo delays. Oscillatory responses to pulses and evoked responses to echoes overlap extensively at short echo delays (5-7 ms), creating interference-like patterns. At longer echo delays, responses are separately evident to both pulses and echoes, with less overlap. These results extend Feng's reports of IC oscillations, and point to different processing mechanisms underlying perception of short vs long echo delays.

Biosonar discrimination of fine surface textures by echolocating free-tailed bats

Echolocating bats are able to discriminate between different surface textures based on the spectral properties of returning echoes. This capability is likely to be important for recognizing prey and for finding suitably perching sites along smooth cave walls. Previous studies showed that bats may exploit echo spectral interference patterns in returning echoes to classify surface textures, but a systematic assessment of the limits of their discrimination performance is lacking and may provide important clues about the neural mechanisms by which bats reconstruct target features based on echo acoustic cues. We trained three Mexican free-tailed bats (Tadarida brasiliensis) on a Y-maze to discriminate between the surfaces of 10 different sheets of aluminum-oxide abrasive sandpapers differing in standardized grit sizes ranging from 40 grit (coarse, 425 μm mean particle diameter) to 240 grit (fine, 54 μm mean particle diameter). Bats were rewarded for choosing the coarsest of two choices. All three bats easily discriminated all abrasive surfaces from a smooth plexiglass control and between all sandpaper comparisons except the two with the smallest absolute difference in mean particle sizes, the 150 vs. 180 grit (92 vs. 82 μm) and the 220 vs. 240 grit (68 vs. 54 μm) surfaces. These results indicate that echolocating free-tailed bats can use slight variations in the echo spectral envelope to remotely classify very fine surface textures with an acuity of at least 23 μm, which rivals direct tactile discrimination performance of the human hand.

Brain-inspired sensorimotor echolocation system for confident landmark recognition

A landmark is a familiar target in terms of the echoes that it can produce and is important for echolocation-based navigation by bats, robots, and blind humans. A brain-inspired system (BIS) achieves confident recognition, defined as classification to an arbitrarily small error probability (PE), by employing a voting process with an echo sequence. The BIS contains sensory neurons implemented with binary single-layer perceptrons trained to classify echo spectrograms with PE and generate excitatory and inhibitory votes in face neurons until a landmark-specific face neuron achieves recognition by reaching a confidence vote level (CVL). A discrete random step process models the vote count to show the recognition probability can achieve any desired accuracy by decreasing PE or increasing CVL. A hierarchical approach first classifies surface reflector and volume scatterer target categories and then uses that result to classify two subcategories that form four landmarks. The BIS models blind human echolocation to recognize four human-made and foliage landmarks by acquiring suitably sized and dense audible echo sequences. The sensorimotor BIS employs landmark-specific CVL values and a 2.7° view increment to acquire echo sequences that achieve zero-error recognition of each landmark independent of the initial view.

Classification of simulated complex echoes based on highlight time separation in the bottlenose dolphin (Tursiops truncatus)

Previous studies suggested that dolphins perceive echo spectral features on coarse (macrospectrum) and fine (microspectrum) scales. This study was based on a finding that these auditory percepts are, to some degree, dependent on the dolphin's ∼250-μs auditory temporal window (i.e., "critical interval"). Here, two dolphins were trained to respond on passively detecting a simulated "target" echo complex [a pair of echo "highlights" with a characteristic 120-μs inter-highlight interval (IHI)]. This target had unique micro- and macrospectral features and was presented among "distractor" echoes with IHIs from 50 to 500 μs (i.e., microspectra) and various highlight durations (i.e., macrospectra). Following acquisition of this discrimination task, probe echo complexes with the macrospectrum of the target but IHIs matching the distractors were infrequently presented. Both dolphins initially responded more often to probes with IHIs of 80-200 μs. Response strategies diverged with increasing probe presentations; one dolphin responded to a progressively narrower range of probe IHIs while the second increased response rates for probes with IHIs > 250 μs. These results support previous conclusions that perception of macrospectra for complex echoes is nonconstant as the IHI decreases below ∼100 μs, but results approaching and exceeding 250 μs-the temporal window upper boundary-were more ambiguous.

Non-invasive auditory brainstem responses to FM sweeps in awake big brown bats

We introduce two EEG techniques, one based on conventional monopolar electrodes and one based on a novel tripolar electrode, to record for the first time auditory brainstem responses (ABRs) from the scalp of unanesthetized, unrestrained big brown bats. Stimuli were frequency-modulated (FM) sweeps varying in sweep direction, sweep duration, and harmonic structure. As expected from previous invasive ABR recordings, upward-sweeping FM signals evoked larger amplitude responses (peak-to-trough amplitude in the latency range of 3–5 ms post-stimulus onset) than downward-sweeping FM signals. Scalp-recorded responses displayed amplitude-latency trading effects as expected from invasive recordings. These two findings validate the reliability of our noninvasive recording techniques. The feasibility of recording noninvasively in unanesthetized, unrestrained bats will energize future research uncovering electrophysiological signatures of perceptual and cognitive processing of biosonar signals in these animals, and allows for better comparison with ABR data from echolocating cetaceans, where invasive experiments are heavily restricted.

Analysis of echolocation behavior of bats in “echo space” using acoustic simulation

Background

Echolocating bats use echo information to perceive space, control their behavior, and adjust flight navigation strategies in various environments. However, the echolocation behavior of bats, including echo information, has not been thoroughly investigated as it is technically difficult to measure all the echoes that reach the bats during flight, even with the conventional telemetry microphones currently in use. Therefore, we attempted to reproduce the echoes received at the location of bats during flight by combining acoustic simulation and behavioral experiments with acoustic measurements. By using acoustic simulation, echoes can be reproduced as temporal waveforms (including diffracted waves and multiple reflections), and detailed echo analysis is possible even in complex obstacle environments.

Results

We visualized the spatiotemporal changes in the echo incidence points detected by bats during flight, which enabled us to investigate the “echo space” revealed through echolocation for the first time. We then hypothesized that by observing the differences in the “echo space” before and after spatial learning, the bats’ attentional position would change. To test this hypothesis, we examined how the distribution of visualized echoes concentrated at the obstacle edges after the bats became more familiar with their environment. The echo incidence points appeared near the edge even when the pulse direction was not toward the edge. Furthermore, it was found that the echo direction correlated with the turn rate of the bat’s flight path, revealing for the first time the relationship between the echo direction and the bat’s flight path.

Conclusions

We were able to clarify for the first time how echoes space affects echolocation behavior in bats by combining acoustic simulations and behavioral experiments.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12915-022-01253-y.

Amplitude discrimination is predictably affected by echo frequency filtering in wideband echolocating bats

Big brown bats echolocate using wideband frequency-modulated (FM) ultrasonic pulses, perceiving target range from echo delay and target size from echo amplitude. Echolocation pulses contain two prominent down-sweeping harmonics (FM1, ∼55-22 kHz; FM2, ∼100-55 kHz), which are affected differently by propagation to the target and back to the bat. Previous work demonstrates that big brown bats utilize the low frequencies in FM1 for target ranging, while FM2 only contributes if FM1 is also present. The present experiments test the hypothesis that the bat's ability to discriminate echo amplitude is also affected by selectively attenuating FM1 or FM2 in target or nontarget echoes. Bats were trained to perform an amplitude discrimination task with virtual echo targets located 83 cm away. Echo delay was fixed and echo amplitude was varied, while either FM1 or FM2 was attenuated by highpass or lowpass filtering. Bats' performance decreased when lower frequencies were attenuated in target echoes and when higher frequencies were attenuated in nontarget echoes. Performance was reversed in the opposite filtering conditions. The bat's ability to distinguish between virtual targets varying in amplitude at the same simulated range indicates a high level of focused attention for perceptual isolation of target from non-target echoes.

Population registration of echo flow in the big brown bat's auditory midbrain

Echolocating big brown bats (Eptesicus fuscus) perceive their surroundings by broadcasting frequency-modulated (FM) ultrasonic pulses and processing returning echoes. Bats echolocate in acoustically cluttered environments containing multiple objects, where each broadcast is followed by multiple echoes at varying time delays. The bat must decipher this complex echo cascade to form a coherent picture of the entire acoustic scene. Neurons in the bat's inferior colliculus (IC) are selective for specific acoustic features of echoes and time delays between broadcasts and echoes. Because of this selectivity, different subpopulations of neurons are activated as the bat flies through its environment, while the physical scene itself remains unchanging. We asked how a neural representation based on variable single-neuron responses could underlie a cohesive perceptual representation of a complex scene. We recorded local field potentials from the IC of big brown bats to examine population coding of echo cascades similar to what the bat might encounter when flying alongside vegetation. We found that the temporal patterning of a simulated broadcast followed by an echo cascade is faithfully reproduced in the population response at multiple stimulus amplitudes and echo delays. Local field potentials to broadcasts and echo cascades undergo amplitude-latency trading consistent with single-neuron data but rarely show paradoxical latency shifts. Population responses to the entire echo cascade move as a unit coherently in time as broadcast-echo cascade delay changes, suggesting that these responses serve as an index for the formation of a cohesive perceptual representation of an acoustic scene.NEW & NOTEWORTHY Echolocating bats navigate through cluttered environments that return cascades of echoes in response to the bat's broadcasts. We show that local field potentials from the big brown bat's auditory midbrain have consistent responses to a simulated echo cascade varying across echo delays and stimulus amplitudes, despite different underlying individual neuronal selectivities. These results suggest that population activity in the midbrain can build a cohesive percept of an auditory scene by aggregating activity over neuronal subpopulations.

Most favorable stimulation duration in the sensorimotor cortex for fNIRS-based BCI

One of the primary objectives of the brain-computer interface (BCI) is to obtain a command with higher classification accuracy within the shortest possible time duration. Therefore, this study evaluates several stimulation durations to propose a duration that can yield the highest classification accuracy. Furthermore, this study aims to address the inherent delay in the hemodynamic responses (HRs) for the command generation time. To this end, HRs in the sensorimotor cortex were evaluated for the functional near-infrared spectroscopy (fNIRS)-based BCI. To evoke brain activity, right-hand-index finger poking and tapping tasks were used. In this study, six different stimulation durations (i.e., 1, 3, 5, 7, 10, and 15 s) were tested on 10 healthy male subjects. Upon stimulation, different temporal features and multiple time windows were utilized to extract temporal features. The extracted features were then classified using linear discriminant analysis. The classification results using the main HR showed that a 5 s stimulation duration could yield the highest classification accuracy, i.e., 74%, with a combination of the mean and maximum value features. However, the results were not significantly different from the classification accuracy obtained using the 15 s stimulation. To further validate the results, a classification using the initial dip was performed. The results obtained endorsed the finding with an average classification accuracy of 73.5% using the features of minimum peak and skewness in the 5 s window. The results based on classification using the initial dip for 5 s were significantly different from all other tested stimulation durations (p < 0.05) for all feature combinations. Moreover, from the visual inspection of the HRs, it is observed that the initial dip occurred as soon as the task started, but the main HR had a delay of more than 2 s. Another interesting finding is that impulsive stimulation in the sensorimotor cortex can result in the generation of a clearer initial dip phenomenon. The results reveal that the command for the fNIRS-based BCI can be generated using the 5 s stimulation duration. In conclusion, the use of the initial dip can reduce the time taken for the generation of commands and can be used to achieve a higher classification accuracy for the fNIRS-BCI within a 5 s task duration rather than relying on longer durations.