Learning auditory discriminations from observation is efficient but less robust than learning from experience

Narula, Gagan; Herbst, Joshua A.; Rychen, Joerg; Hahnloser, Richard H. R.

doi:10.1038/s41467-018-05422-y

Cited by 17 publications

(13 citation statements)

References 37 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Depending on task requirements, these constraints may lead to compressed representations of the observed past. Compressed representations are also associated behaviorally [4][5][6][7] and theoretically [8][9][10] with more efficient learning and generalization. Compressed representations currently cannot be measured directly in the human brain, but nevertheless can still impact the electrophysiological responses to each stimulus in a sequence as a function of the representation of preceding stimuli (for a similar methodology see [11]).…”

Section: Introductionmentioning

confidence: 99%

Surprise response as a probe for compressed memory states

2020

View full text Add to dashboard Cite

The limited capacity of recent memory inevitably leads to partial memory of past stimuli. There is also evidence that behavioral and neural responses to novel or rare stimuli are dependent on one's memory of past stimuli. Thus, these responses may serve as a probe of different individuals' remembering and forgetting characteristics. Here, we utilize two lossy compression models of stimulus sequences that inherently involve forgetting, which in addition to being a necessity under many conditions, also has theoretical and behavioral advantages. One model is based on a simple stimulus counter and the other on the Information Bottleneck (IB) framework which suggests a more general, theoretically justifiable principle for biological and cognitive phenomena. These models are applied to analyze a novelty-detection event-related potential commonly known as the P300. The trial-by-trial variations of the P300 response, recorded in an auditory oddball paradigm, were subjected to each model to extract two stimulus-compression parameters for each subject: memory length and representation accuracy. These parameters were then utilized to estimate the subjects' recent memory capacity limit under the task conditions. The results, along with recently published findings on single neurons and the IB model, underscore how a lossy compression framework can be utilized to account for trial-by-trial variability of neural responses at different spatial scales and in different individuals, while at the same time providing estimates of individual memory characteristics at different levels of representation using a theoretically-based parsimonious model. Author summarySurprise responses reflect expectations based on preceding stimuli representations, and hence can be used to infer the characteristics of memory utilized for a task. We suggest a quantitative method for extracting an individual estimate of effective memory capacity dedicated for a task based on the correspondence between a theoretical surprise model and electrophysiological single-trial surprise responses. We demonstrate this method on EEG responses recorded while participants were performing a simple auditory task; we PLOS Computational Biology | https://doi.

show abstract

Section: Introductionmentioning

confidence: 99%

Surprise response as a probe for compressed memory states

2020

View full text Add to dashboard Cite

show abstract

“…While learning through visual exposure (observation) has been well studied 22,[29][30][31] , the acquisition of complex behaviors typically depends on multiple sensory cues from the environment and from social interactions. For instance, juvenile songbirds can learn elements of a species-specific vocalization through auditory exposure alone, but learning is enhanced by the presence of an adult tutor [32][33][34][35][36][37] . Similarly, filial imprinting involves many sensory and social cues 38 , but exposure to an auditory cue alone is sufficient to maintain or induce the behavior 39,40 .…”

mentioning

confidence: 99%

Social learning exploits the available auditory or visual cues

Paraouty

Charbonneau

Sanes

2020

Sci Rep

View full text Add to dashboard Cite

to learn in a social context. One plausible hypothesis that explains our results is that naïve gerbils can learn about the spatial behavior of demonstrators, either through visual or auditory signals, and can subsequently deploy this information when exploring and training on the auditory discrimination task. Thus, social learning is flexible and may, in part, explain its ubiquitous presence and function in many species across an array of environments. Methods Experimental animals. Gerbil (Meriones unguiculatus, n = 118) pups were weaned at postnatal day (P) 30 from commercial breeding pairs (Charles River). Littermates were caged together, but separated by sex, and maintained in a 12 h light/dark cycle. Same-sex litters were split into 2 subsets: 1. experimenter-trained demonstrator gerbils, and 2. naïve gerbils for exposure and practice. Each naïve animal was used for one single experimental or control paradigm. Data for each experimental or control paradigm was collected from at least 3 different litters in order to avoid any biases. Finally, animals in a given litter were used for at least 2 paradigms. All procedures related to the maintenance and use of animals were approved by the University Animal Welfare Committee at New York University, and all experiments were performed in accordance with the relevant guidelines and regulations.

show abstract

“…Echo attenuation has in the past enabled studies of the effect of communication on the behavior of individuals within their isolated environments. A two-channel version of the system presented here has enabled targeted control of auditory input in a study of learning from observation 19 . In that study, a demonstrator bird was separated from an observer bird; while the demonstrator was engaged in learning an auditory discrimination task from aversive reinforcement, the observer could gain information about the auditory stimuli and the demonstrator’s behavior.…”

Section: Discussionmentioning

confidence: 99%

A system for controlling vocal communication networks

Rychen

Rodrigues

Tomka

et al. 2021

Sci Rep

View full text Add to dashboard Cite

Animal vocalizations serve a wide range of functions including territorial defense, courtship, social cohesion, begging, and vocal learning. Whereas many insights have been gained from observational studies and experiments using auditory stimulation, there is currently no technology available for the selective control of vocal communication in small animal groups. We developed a system for real-time control of vocal interactions among separately housed animals. The system is implemented on a field-programmable gate array (FPGA) and it allows imposing arbitrary communication networks among up to four animals. To minimize undesired transitive sound leakage, we adopted echo attenuation and sound squelching algorithms. In groups of three zebra finches, we restrict vocal communication in circular and in hierarchical networks and thereby mimic complex eavesdropping and middleman situations.

show abstract

Learning auditory discriminations from observation is efficient but less robust than learning from experience

Cited by 17 publications

References 37 publications

Surprise response as a probe for compressed memory states

Surprise response as a probe for compressed memory states

Social learning exploits the available auditory or visual cues

A system for controlling vocal communication networks

Contact Info

Product

Resources

About