Limits...
Probing real sensory worlds of receivers with unsupervised clustering.

Pfeiffer M, Hartbauer M, Lang AB, Maass W, Römer H - PLoS ONE (2012)

Bottom Line: Our results show that the reliability of burst coding in the time domain is so high that identical stimuli lead to extremely similar spike pattern responses, even for different preparations on different dates, and even if one of the preparations is recorded outdoors and the other one in the sound proof lab.Our study shows that burst coding can provide a reliable mechanism for acoustic insects to classify and discriminate signals under very noisy real-world conditions.This gives new insights into the neural mechanisms potentially used by bushcrickets to discriminate conspecific songs from sounds of predators in similar carrier frequency bands.

View Article: PubMed Central - PubMed

Affiliation: Institute for Theoretical Computer Science, TU Graz, Graz, Austria. pfeiffer@ini.phys.ethz.ch

ABSTRACT
The task of an organism to extract information about the external environment from sensory signals is based entirely on the analysis of ongoing afferent spike activity provided by the sense organs. We investigate the processing of auditory stimuli by an acoustic interneuron of insects. In contrast to most previous work we do this by using stimuli and neurophysiological recordings directly in the nocturnal tropical rainforest, where the insect communicates. Different from typical recordings in sound proof laboratories, strong environmental noise from multiple sound sources interferes with the perception of acoustic signals in these realistic scenarios. We apply a recently developed unsupervised machine learning algorithm based on probabilistic inference to find frequently occurring firing patterns in the response of the acoustic interneuron. We can thus ask how much information the central nervous system of the receiver can extract from bursts without ever being told which type and which variants of bursts are characteristic for particular stimuli. Our results show that the reliability of burst coding in the time domain is so high that identical stimuli lead to extremely similar spike pattern responses, even for different preparations on different dates, and even if one of the preparations is recorded outdoors and the other one in the sound proof lab. Simultaneous recordings in two preparations exposed to the same acoustic environment reveal that characteristics of burst patterns are largely preserved among individuals of the same species. Our study shows that burst coding can provide a reliable mechanism for acoustic insects to classify and discriminate signals under very noisy real-world conditions. This gives new insights into the neural mechanisms potentially used by bushcrickets to discriminate conspecific songs from sounds of predators in similar carrier frequency bands.

Show MeSH
Analysis of bursts extracted from the spike data.A) The joint interspike-interval plot for a single preparation indicates the presence of bursts by a large cluster of points in the lower left corner, which represents periods of fast firing, and clusters of points in the upper left and lower right corner, which indicate onsets and offset of bursts. B) Histogram of inter-burst intervals (bin size: 0.5 s). C) 13 bursts extracted from 1 minute of the recording. The set of bursts contains 2 responses to stimulus 3 (bursts 10 and 13), 2 responses to stimulus 4 (bursts 4 and 8), and 9 responses to different sources of environmental noise. D) Detection of bursts in spike trains. The 6 spikes in the shaded area constitute a burst, because they are separated by time window of at least 60 ms from the first spike, the interspike-interval is never larger than 30 ms, the burst duration is longer than 8 ms and there are more than 5 spikes.
© Copyright Policy
Related In: Results  -  Collection


getmorefigures.php?uid=PMC3368931&req=5

pone-0037354-g002: Analysis of bursts extracted from the spike data.A) The joint interspike-interval plot for a single preparation indicates the presence of bursts by a large cluster of points in the lower left corner, which represents periods of fast firing, and clusters of points in the upper left and lower right corner, which indicate onsets and offset of bursts. B) Histogram of inter-burst intervals (bin size: 0.5 s). C) 13 bursts extracted from 1 minute of the recording. The set of bursts contains 2 responses to stimulus 3 (bursts 10 and 13), 2 responses to stimulus 4 (bursts 4 and 8), and 9 responses to different sources of environmental noise. D) Detection of bursts in spike trains. The 6 spikes in the shaded area constitute a burst, because they are separated by time window of at least 60 ms from the first spike, the interspike-interval is never larger than 30 ms, the burst duration is longer than 8 ms and there are more than 5 spikes.

Mentions: Artificial acoustic signals, as well as some natural background stimuli caused bursting activity in the nerve cell, i.e. it was firing at a much increased rate compared to its baseline firing activity. We extracted from the continuous recordings in natural habitats all these short time segments in which the omega neuron was bursting. Our criterion for detecting bursts in a continuous stream of spikes required a silent interval of at least 60 ms before the start of the burst, a constantly high firing rate of at least 33 Hz, a minimum duration of 8 ms, and a minimum of 5 spikes within the burst (see Figure 2D, as well as Burst Detection in Materials and Methods for more details). In Figure 2 we illustrate the analysis of one recording session. From the joint interspike-interval (ISI) diagram in Figure 2A, which shows the duration of the next ISI as a function of the preceding ISI, one can see the presence of bursts in the recordings. By definition a burst is a period of rapid firing, preceded and followed by a longer period of no or low activity. The accumulation of points in the lower left corner indicates that there are numerous periods of rapid firing, which are typical for firing intervals within bursts. The clusters of points in the upper left and lower right corner show that there are also many short ISIs preceded or followed by longer intervals, which indicate the onsets or offsets of bursts. The intervals between bursts display no clear pattern, but the histogram of inter-burst intervals in Figure 2B shows that most intervals are short, and the frequency of longer inter-burst intervals decays. In Figure 2C we plotted the bursts contained in one minute of recordings in the original order in which they appeared. Looking only at the raw data it is not immediately clear which bursts belong to a common cluster, and although the bursts are from relatively close time points, there is no visible structure of bursts in response to environmental noise.


Probing real sensory worlds of receivers with unsupervised clustering.

Pfeiffer M, Hartbauer M, Lang AB, Maass W, Römer H - PLoS ONE (2012)

Analysis of bursts extracted from the spike data.A) The joint interspike-interval plot for a single preparation indicates the presence of bursts by a large cluster of points in the lower left corner, which represents periods of fast firing, and clusters of points in the upper left and lower right corner, which indicate onsets and offset of bursts. B) Histogram of inter-burst intervals (bin size: 0.5 s). C) 13 bursts extracted from 1 minute of the recording. The set of bursts contains 2 responses to stimulus 3 (bursts 10 and 13), 2 responses to stimulus 4 (bursts 4 and 8), and 9 responses to different sources of environmental noise. D) Detection of bursts in spike trains. The 6 spikes in the shaded area constitute a burst, because they are separated by time window of at least 60 ms from the first spike, the interspike-interval is never larger than 30 ms, the burst duration is longer than 8 ms and there are more than 5 spikes.
© Copyright Policy
Related In: Results  -  Collection

Show All Figures
getmorefigures.php?uid=PMC3368931&req=5

pone-0037354-g002: Analysis of bursts extracted from the spike data.A) The joint interspike-interval plot for a single preparation indicates the presence of bursts by a large cluster of points in the lower left corner, which represents periods of fast firing, and clusters of points in the upper left and lower right corner, which indicate onsets and offset of bursts. B) Histogram of inter-burst intervals (bin size: 0.5 s). C) 13 bursts extracted from 1 minute of the recording. The set of bursts contains 2 responses to stimulus 3 (bursts 10 and 13), 2 responses to stimulus 4 (bursts 4 and 8), and 9 responses to different sources of environmental noise. D) Detection of bursts in spike trains. The 6 spikes in the shaded area constitute a burst, because they are separated by time window of at least 60 ms from the first spike, the interspike-interval is never larger than 30 ms, the burst duration is longer than 8 ms and there are more than 5 spikes.
Mentions: Artificial acoustic signals, as well as some natural background stimuli caused bursting activity in the nerve cell, i.e. it was firing at a much increased rate compared to its baseline firing activity. We extracted from the continuous recordings in natural habitats all these short time segments in which the omega neuron was bursting. Our criterion for detecting bursts in a continuous stream of spikes required a silent interval of at least 60 ms before the start of the burst, a constantly high firing rate of at least 33 Hz, a minimum duration of 8 ms, and a minimum of 5 spikes within the burst (see Figure 2D, as well as Burst Detection in Materials and Methods for more details). In Figure 2 we illustrate the analysis of one recording session. From the joint interspike-interval (ISI) diagram in Figure 2A, which shows the duration of the next ISI as a function of the preceding ISI, one can see the presence of bursts in the recordings. By definition a burst is a period of rapid firing, preceded and followed by a longer period of no or low activity. The accumulation of points in the lower left corner indicates that there are numerous periods of rapid firing, which are typical for firing intervals within bursts. The clusters of points in the upper left and lower right corner show that there are also many short ISIs preceded or followed by longer intervals, which indicate the onsets or offsets of bursts. The intervals between bursts display no clear pattern, but the histogram of inter-burst intervals in Figure 2B shows that most intervals are short, and the frequency of longer inter-burst intervals decays. In Figure 2C we plotted the bursts contained in one minute of recordings in the original order in which they appeared. Looking only at the raw data it is not immediately clear which bursts belong to a common cluster, and although the bursts are from relatively close time points, there is no visible structure of bursts in response to environmental noise.

Bottom Line: Our results show that the reliability of burst coding in the time domain is so high that identical stimuli lead to extremely similar spike pattern responses, even for different preparations on different dates, and even if one of the preparations is recorded outdoors and the other one in the sound proof lab.Our study shows that burst coding can provide a reliable mechanism for acoustic insects to classify and discriminate signals under very noisy real-world conditions.This gives new insights into the neural mechanisms potentially used by bushcrickets to discriminate conspecific songs from sounds of predators in similar carrier frequency bands.

View Article: PubMed Central - PubMed

Affiliation: Institute for Theoretical Computer Science, TU Graz, Graz, Austria. pfeiffer@ini.phys.ethz.ch

ABSTRACT
The task of an organism to extract information about the external environment from sensory signals is based entirely on the analysis of ongoing afferent spike activity provided by the sense organs. We investigate the processing of auditory stimuli by an acoustic interneuron of insects. In contrast to most previous work we do this by using stimuli and neurophysiological recordings directly in the nocturnal tropical rainforest, where the insect communicates. Different from typical recordings in sound proof laboratories, strong environmental noise from multiple sound sources interferes with the perception of acoustic signals in these realistic scenarios. We apply a recently developed unsupervised machine learning algorithm based on probabilistic inference to find frequently occurring firing patterns in the response of the acoustic interneuron. We can thus ask how much information the central nervous system of the receiver can extract from bursts without ever being told which type and which variants of bursts are characteristic for particular stimuli. Our results show that the reliability of burst coding in the time domain is so high that identical stimuli lead to extremely similar spike pattern responses, even for different preparations on different dates, and even if one of the preparations is recorded outdoors and the other one in the sound proof lab. Simultaneous recordings in two preparations exposed to the same acoustic environment reveal that characteristics of burst patterns are largely preserved among individuals of the same species. Our study shows that burst coding can provide a reliable mechanism for acoustic insects to classify and discriminate signals under very noisy real-world conditions. This gives new insights into the neural mechanisms potentially used by bushcrickets to discriminate conspecific songs from sounds of predators in similar carrier frequency bands.

Show MeSH