Cortical signatures of auditory looming bias show cue-specific adaptation between newborns and young adults

Abstract

Adaptive biases in favor of approaching, or "looming", sounds have been found across ages and species, thereby implicating the potential of their evolutionary origin and universal basis. The human auditory system is well-developed at birth, yet spatial hearing abilities further develop with age. To disentangle the speculated inborn, evolutionary component of the auditory looming bias from its learned counterpart, we collected high-density electroencephalographic data across human adults and newborns. As distance-motion cues we manipulated either the sound's intensity or spectral shape, which is pinna-induced and thus prenatally inaccessible. Through cortical source localisation we demonstrated the emergence of the bias in both age groups at the level of Heschl's gyrus. Adults exhibited the bias in both attentive and inattentive states; yet differences in amplitude and latency appeared based on attention and cue type. Contrary to the adults, in newborns the bias was elicited only through manipulations of intensity and not spectral cues. We conclude that the looming bias comprises innate components while flexibly incorporating the spatial cues acquired through lifelong exposure.

Keywords: Cortex; Human behaviour.

Fig. 1. Experimental design.

a Illustration of experimental factors movement and cue type. The transition between two sounds of different intensities (top, blue) or spectral shapes (bottom, red) creates the sensation of a moving sound source. Thick arrows represent a 50% transition probability for motion trials (dark = looming; light = receding), while thin circular arrows indicate a 50% probability for static trials. b Magnitude profile over time (top panel) and frequency (bottom panel) of all implemented stimuli. Filtering by the native spectral shape evokes a spatially externalized auditory percept. Sounds devoid of native spectral characteristics (flat spectrum) do not elicit this externalization, making sounds appear close to one’s ear.

Fig. 2. Model-based analysis of adults’ behavioral responses indicates speed-up of evidence accumulation for looming sounds.

a Response times and accuracies contrasted between actual data and simulated responses generated by a linear ballistic accumulator model with fitted group-level parameters. Symbols denote means for response times and medians for accuracies. Error bars denote the standard deviation for response times and the first and third quartiles for accuracies. b Posterior distributions of drift rate estimates indicating the listeners' speed of evidence accumulation for correctly discriminated motion directions. Center lines show medians, box limits show interquartile ranges, and whiskers show ranges up to 1.5 times the interquartile range. N = 28.

Fig. 3. Adults’ change-evoked scalp potentials reveal auditory looming bias across attentional states and cue types.

a Potentials evoked at the vertex electrode (Cz) on average across looming and receding trials (looming/2 + receding/2). Shaded areas denote the standard errors of the means. b Difference waveforms (looming − receding) at the vertex electrode. c Extracted peak amplitude values of the N1 and P2 components. Error bars represent 95% confidence intervals. Asterisks indicate significant main effects (p < 0.05) per component. d Extracted peak latency values of the N1 and P2 components. e Scalp topographies and duration of clusters with significant looming bias, defined as the difference between looming and receding trials. Horizontal lines denote the durations of the significant clusters and are tagged with numbers at the point of maximum manifestation. N = 28.

Fig. 4. Adult participants’ ERPs locked to sound onset show no differences between near and far distances.

a Grand-average topographic maps around N1 and P2 deflections (top) and evoked Cz potentials (bottom) depending on attention, averaged over cue type and distance. b Comparisons of evoked Cz potentials between distances within cue type. Shaded areas denote standard errors of means N = 28.

Fig. 5. Change-evoked activity in HG of the adult participants reveals auditory looming bias across attentional states and cue types.

a Evoked activity, averaged over looming and receding trials, for left HG (left figure column) and right HG (right column), including lateral views of whole-brain source activations at 120 ms (N1 peak). Blue contours within the brain maps indicate the borders of the target ROIs (HG). Shaded areas depict the standard errors of the means. b Looming bias (looming − receding) evoked activity for left HG (left) and right HG (right). Horizontal lines denote the durations of significant temporal clusters. c Peak N1 and P2 amplitude values for evoked HG activity depending on brain hemisphere, type of cue (intensity/spectral shape changes), and attentional state (active/passive). Error bars represent 95% confidence intervals. Asterisks indicate significant main effects (p < 0.05) per component. d Peak N1 and P2 latency values for evoked HG activity. N = 28.

Fig. 6. Change-evoked scalp potentials from newborns reveal auditory looming bias only for the intensity condition.

a Responses at the frontocentral electrode cluster for different cue types averaged across looming and receding trials. Shaded areas denote the standard errors of the means. b Looming versus receding neural responses for the change-locked intensity cue condition. The gray bar denotes the duration of the significant looming bias. c Looming versus receding neural responses for the change-locked spectral cue condition. d Topographic analysis of looming bias elicited by intensity cues. N = 71.

Fig. 7. Newborns’ onset-evoked scalp potentials at the defined electrode cluster reveal no auditory position bias for any of the considered starting positions.

a Onset potentials averaged across trials of near and far positions. b Near versus far neural responses for the onset-locked intensity condition. c Near versus far neural responses for the onset-locked spectral condition. Shaded areas denote the standard errors of the means (N = 71).

Fig. 8. Change-evoked HG activity in newborns reveals auditory looming bias only for the intensity condition.

a Brainmaps pooled across conditions and averaged within the time interval 250–300 ms. b, c Activity evoked by looming vs. receding sounds in left (left panels) and right (right panels) HG based on intensity (b) and spectral (c) cues. Gray areas denote the duration of significant temporal clusters. Shaded areas denote the standard errors of the means. N = 71.