Adaptation supports short-term memory in a visual change detection task

Hu, Brian; Garrett, Marina; Groblewski, Peter A.; Ollerenshaw, Douglas R.; Shang, Jiaqi; Roll, Kate; Manavi, Sahar; Koch, Christof; Olsen, Shawn R.; Mihalaş, Ştefan

doi:10.1101/2020.03.06.977512

Cited by 4 publications

(5 citation statements)

References 33 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…STP-D of synapses, also referred to as paired-pulse depression, reduces the amplitude of postsynaptic potentials for later spikes in a spike train. The impact of this mechanism on simple temporal computing tasks had been examined in a number of publications ( Maass et al, 2002 ; Buonomano and Maass, 2009 ; Masse et al, 2019 ; Hu et al, 2020 ).…”

Section: Resultsmentioning

confidence: 99%

“…However, for a more demanding time-series classification task short-term depression of synapses provides about the same performance as SFA (see Figure 2C ). An important contribution of depressing synapses for temporal computing has already previously been proposed ( Hu et al, 2020 ). This is on first sight counter-intuitive, just as the fact that spike-triggered reduction rather than increase of neural excitability provides better support for temporal computing.…”

Section: Discussionmentioning

confidence: 99%

See 1 more Smart Citation

Spike frequency adaptation supports network computations on temporally dispersed information

Salaj

Subramoney

Kraišniković

et al. 2021

eLife

View full text Add to dashboard Cite

For solving tasks such as recognizing a song, answering a question, or inverting a sequence of symbols, cortical microcircuits need to integrate and manipulate information that was dispersed over time during the preceding seconds. Creating biologically realistic models for the underlying computations, especially with spiking neurons and for behaviorally relevant integration time spans, is notoriously difficult. We examine the role of spike frequency adaptation in such computations and find that it has a surprisingly large impact. The inclusion of this well-known property of a substantial fraction of neurons in the neocortex – especially in higher areas of the human neocortex – moves the performance of spiking neural network models for computations on network inputs that are temporally dispersed from a fairly low level up to the performance level of the human brain.

show abstract

Section: Resultsmentioning

confidence: 99%

Section: Discussionmentioning

confidence: 99%

Spike frequency adaptation supports network computations on temporally dispersed information

Salaj

Subramoney

Kraišniković

et al. 2021

eLife

View full text Add to dashboard Cite

show abstract

“…An important contribution of depressing synapses for temporal computing has already previously been proposed (Hu et al, 2020). This is on first sight counter-intuitive, just as the fact that spike-triggered reduction rather than increase of neural excitability provides better support for temporal computing.…”

Section: Discussionmentioning

confidence: 99%

“…Depressing short-term plasticity of synapses, also referred to as paired-pulse depression, reduces the amplitude of postsynaptic potentials for later spikes in a spike train. The impact of this mechanism on simple temporal computing tasks had been examined in a number of publications (Maass et al, 2002;Buonomano and Maass, 2009;Masse et al, 2019;Hu et al, 2020).…”

Section: Mation and Yields Powerful Generalization Capabilitymentioning

confidence: 99%

Spike frequency adaptation supports network computations on temporally dispersed information

Salaj

Subramoney

Kraišniković

et al. 2020

Preprint

View full text Add to dashboard Cite

12Brains are able to integrate memory from the recent past into their current computations, 13 seemingly without effort. This ability is critical for cognitive tasks such as speech understand-14 ing or working with sequences of symbols according to dynamically changing rules. But it has 15 remained unknown how networks of spiking neurons in the brain can achieve that. We show 16 that the presence of neurons with spike frequency adaptation makes a significant difference: 17 Their inclusion in a network moves its performance for such computing tasks from a very low 18 level close to the level of human performance. While artificial neural networks with special 19 long short-term memory (LSTM) units had already reached such high performance levels, 20 they lack biological plausibility. We find that neurons with spike-frequency adaptation, which 21 occur especially frequently in higher cortical areas of the human brain, provide to brains a 22 functional equivalent to LSTM units. 23 Introduction 24Our brains are able to constantly process new information in the light of recent experiences and 25 dynamically changing rules, seemingly without any effort. But we do not know how networks of 26 spiking neurons (SNNs) in the brain accomplish that. The performance of both spike-based and 27 rate-based models for recurrent neural networks in the brain have stayed on a rather low perfor-28 mance level for such tasks, far below the performance level of the human brain and artificial neural 29 network models. Artificial neural network models that perform well on such tasks use, instead of 30 neuron models, a special type of unit called Long Short-Term Memory (LSTM) unit. LSTM units 31 store information in registers -like a digital computer -where it remains without perturbance 32 by network activity for an indefinite amount of time, until is is actively updated or recalled. Hence 33 these LSTM units are not biologically plausible, and it has remained an open problem how neural 34 networks in the brain achieve so high performance on cognitively demanding tasks that require 35 integration of information from the recent past into current computational processing. We pro-36 pose that the brain achieves this -at least for some tasks -without separating computation and 37 short-term memory in different network modules: Rather it intertwines computing and memory 38 with the help of inherent slow dynamic processes in neurons and synapses. 39 Arguably the most prominent internal dynamics of neurons on the time scale of seconds -40 which is particularly relevant for many cognitive tasks -is spike-frequency adaptation (SFA). It 41 1 is expressed by a substantial fraction of neocortical neurons (Allen Institute, 2018). SFA reduces 42 the excitability of a neuron in response to its firing, see Fig. 1. Neurons with SFA have often 43 been included in SNN models that aim at modelling the dynamics of brain networks (Gutkin and 44 Zeldenrust, 2014), but not in computational studies. We show that neurons with SFA do in fact 45 si...

show abstract

“…Moreover, our task includes features that permit investigation of the physiological correlates of behavior and cognition. For instance, this task allows for exploration of stimulus novelty and learning, temporal expectation (Garrett et al, 2020), and short-term memory (Hu et al, 2020).…”

Section: Introductionmentioning

confidence: 99%

Characterization of Learning, Motivation, and Visual Perception in Five Transgenic Mouse Lines Expressing GCaMP in Distinct Cell Populations

Groblewski

Ollerenshaw

Kiggins

et al. 2020

Front. Behav. Neurosci.

Self Cite

View full text Add to dashboard Cite

To study the mechanisms of perception and cognition, neural measurements must be made during behavior. A goal of the Allen Brain Observatory is to map the activity of distinct cortical cell classes underlying visual and behavioral processing. Here we describe standardized methodology for training head-fixed mice on a visual change detection task, and we use our paradigm to characterize learning and behavior of five GCaMP6-expressing transgenic lines. We used automated training procedures to facilitate comparisons across mice. Training times varied, but most transgenic mice learned the behavioral task. Motivation levels also varied across mice. To compare mice in similar motivational states we subdivided sessions into over-, under-, and optimally motivated periods. When motivated, the pattern of perceptual decisions were highly correlated across transgenic lines, although overall performance (d-prime) was lower in one line labeling somatostatin inhibitory cells. These results provide important context for using these mice to map neural activity underlying perception and behavior.

show abstract

Adaptation supports short-term memory in a visual change detection task

Cited by 4 publications

References 33 publications

Spike frequency adaptation supports network computations on temporally dispersed information

Spike frequency adaptation supports network computations on temporally dispersed information

Spike frequency adaptation supports network computations on temporally dispersed information

Characterization of Learning, Motivation, and Visual Perception in Five Transgenic Mouse Lines Expressing GCaMP in Distinct Cell Populations

Contact Info

Product

Resources

About