Pub Date : 1900-01-01DOI: 10.32470/ccn.2019.1146-0
B. Peters, N. Kriegeskorte
Human perception decomposes the world into represented objects that are selectively attended, tracked, and predicted as we engage our surroundings. Object representations emancipate perception from the senses, enabling us to keep in mind whats out of sight, and provide a stepping stone toward more abstract symbolic cognition. Human behavioral studies have captured cognitive objects by documenting empirical phenomena including object permanence, proto-objects, and object files. Current deep neural network (DNN) models of visual object recognition, by contrast, remain largely tethered to the sensory input — despite achieving human-level performance at labeling objects in images. Here, we review the key behavioral phenomena and cognitive concepts related to perceptual objects. We then link them to early-stage neural network mechanisms that capture certain aspects of these phenomena. We argue that the human behavioral and cognitive literature provides a starting point for experimental paradigms that can not only reveal mechanisms of human cognition, but also serve as benchmarks driving development of a new class of deep neural network models of vision that will put the object into object recognition.
{"title":"What is a perceptual object? Human behavioral challenges for deep neural network modeling","authors":"B. Peters, N. Kriegeskorte","doi":"10.32470/ccn.2019.1146-0","DOIUrl":"https://doi.org/10.32470/ccn.2019.1146-0","url":null,"abstract":"Human perception decomposes the world into represented objects that are selectively attended, tracked, and predicted as we engage our surroundings. Object representations emancipate perception from the senses, enabling us to keep in mind whats out of sight, and provide a stepping stone toward more abstract symbolic cognition. Human behavioral studies have captured cognitive objects by documenting empirical phenomena including object permanence, proto-objects, and object files. Current deep neural network (DNN) models of visual object recognition, by contrast, remain largely tethered to the sensory input — despite achieving human-level performance at labeling objects in images. Here, we review the key behavioral phenomena and cognitive concepts related to perceptual objects. We then link them to early-stage neural network mechanisms that capture certain aspects of these phenomena. We argue that the human behavioral and cognitive literature provides a starting point for experimental paradigms that can not only reveal mechanisms of human cognition, but also serve as benchmarks driving development of a new class of deep neural network models of vision that will put the object into object recognition.","PeriodicalId":281121,"journal":{"name":"2019 Conference on Cognitive Computational Neuroscience","volume":"22 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1900-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123022500","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 1900-01-01DOI: 10.32470/ccn.2019.1010-0
K. Seeliger, L. Ambrogioni, Umut Güçlü, M. Gerven
Neural information flow (NIF) is a new framework for system identification in neuroscience. NIF models represent neural information processing systems as coupled brain regions that each embody neural computations. These brain regions are coupled to observed data specific to that region via linear observation models. NIF models are trained via backpropagation, directly leveraging the neural signal as the loss. Trained NIF models are accessible for in silico analyses. Using a large-scale fMRI video stimulation dataset and a feed-forward convolutional neural network-based NIF model as an example we show that, in this manner, we can estimate models that learn meaningful neural computations and representations. Our framework is general in the sense that it can be used in conjunction with any neural recording techniques. It is also scalable, providing neuroscientists with a principled approach to make sense of high-dimensional neural datasets.
神经信息流(Neural information flow, NIF)是神经科学中一种新的系统识别框架。NIF模型将神经信息处理系统表示为耦合的大脑区域，每个区域都包含神经计算。这些大脑区域通过线性观察模型与特定于该区域的观察数据相关联。NIF模型通过反向传播进行训练，直接利用神经信号作为损失。经过训练的NIF模型可用于计算机分析。以大规模fMRI视频刺激数据集和基于前馈卷积神经网络的NIF模型为例，我们表明，通过这种方式，我们可以估计学习有意义的神经计算和表示的模型。我们的框架在某种意义上是通用的，它可以与任何神经记录技术结合使用。它也是可扩展的，为神经科学家提供了一种有原则的方法来理解高维神经数据集。
{"title":"Neural Information Flow: Learning neural information processing systems from brain activity","authors":"K. Seeliger, L. Ambrogioni, Umut Güçlü, M. Gerven","doi":"10.32470/ccn.2019.1010-0","DOIUrl":"https://doi.org/10.32470/ccn.2019.1010-0","url":null,"abstract":"Neural information flow (NIF) is a new framework for system identification in neuroscience. NIF models represent neural information processing systems as coupled brain regions that each embody neural computations. These brain regions are coupled to observed data specific to that region via linear observation models. NIF models are trained via backpropagation, directly leveraging the neural signal as the loss. Trained NIF models are accessible for in silico analyses. Using a large-scale fMRI video stimulation dataset and a feed-forward convolutional neural network-based NIF model as an example we show that, in this manner, we can estimate models that learn meaningful neural computations and representations. Our framework is general in the sense that it can be used in conjunction with any neural recording techniques. It is also scalable, providing neuroscientists with a principled approach to make sense of high-dimensional neural datasets.","PeriodicalId":281121,"journal":{"name":"2019 Conference on Cognitive Computational Neuroscience","volume":"49 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1900-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131230091","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 1900-01-01DOI: 10.32470/ccn.2019.1067-0
M. T. Wal, J. Domingo, Julia Lifanov, Frederic Roux, Luca D. Kolibius, D. Rollings, V. Sawlani, R. Chelvarajah, B. Staresina, S. Hanslmayr, M. Wimber
{"title":"Oscillatory Patterns in Behavioral Responses during a Memory Task","authors":"M. T. Wal, J. Domingo, Julia Lifanov, Frederic Roux, Luca D. Kolibius, D. Rollings, V. Sawlani, R. Chelvarajah, B. Staresina, S. Hanslmayr, M. Wimber","doi":"10.32470/ccn.2019.1067-0","DOIUrl":"https://doi.org/10.32470/ccn.2019.1067-0","url":null,"abstract":"","PeriodicalId":281121,"journal":{"name":"2019 Conference on Cognitive Computational Neuroscience","volume":"32 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1900-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115172673","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 1900-01-01DOI: 10.32470/ccn.2019.1377-0
O. Zika, K. Wiech, Nicolas W. Schuck
Learning the likelihood of aversive events is achieved either by gradual learning or via inference of hidden states. We previously linked the tendency towards state switching to trait anxiety but the effect of environmental noise has not been investigated. In the present study we employ a Pavlovian probabilistic learning paradigm to test how environmental noise promotes either state switching or gradual lerning. Participants completed three sessions varying in shock contingency jumps (60/40%, 75/25% or 90/10%). As a signature of state-switching we analyzed steepness of post-reversal switch. In support of our hypothesis we found that steepest switches were present in the 90/10 environment. This effect was found to be driven by high trait anxiety. Trait anxiety also positively correlated with difference between acquisition and extinction. Next, we developed a state switching model and performed model comparison using cross-validation. Analysis of model parameters found positive correlation between trait anxiety and tendency to create more states. In summary, our behavioural and modelling result show that less noisy environments encourage state switching, and that anxious individual have an increased tendency to represent the environment as multiple states. This result highlights trait anxiety as vulnerability in successful extinction treatment.
{"title":"High trait anxious individuals represent aversive environment as multiple states: a computational mechanism behind reinstatement","authors":"O. Zika, K. Wiech, Nicolas W. Schuck","doi":"10.32470/ccn.2019.1377-0","DOIUrl":"https://doi.org/10.32470/ccn.2019.1377-0","url":null,"abstract":"Learning the likelihood of aversive events is achieved either by gradual learning or via inference of hidden states. We previously linked the tendency towards state switching to trait anxiety but the effect of environmental noise has not been investigated. In the present study we employ a Pavlovian probabilistic learning paradigm to test how environmental noise promotes either state switching or gradual lerning. Participants completed three sessions varying in shock contingency jumps (60/40%, 75/25% or 90/10%). As a signature of state-switching we analyzed steepness of post-reversal switch. In support of our hypothesis we found that steepest switches were present in the 90/10 environment. This effect was found to be driven by high trait anxiety. Trait anxiety also positively correlated with difference between acquisition and extinction. Next, we developed a state switching model and performed model comparison using cross-validation. Analysis of model parameters found positive correlation between trait anxiety and tendency to create more states. In summary, our behavioural and modelling result show that less noisy environments encourage state switching, and that anxious individual have an increased tendency to represent the environment as multiple states. This result highlights trait anxiety as vulnerability in successful extinction treatment.","PeriodicalId":281121,"journal":{"name":"2019 Conference on Cognitive Computational Neuroscience","volume":"48 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1900-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128135157","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 1900-01-01DOI: 10.32470/ccn.2019.1315-0
Ivan Iotzov, L. Parra
Speech signals have the ability to entrain brain activity to the rapid fluctuations found in speech sounds. This entrainment can be measured using electroencephalographic (EEG) recordings and is strong enough to allow discrimination between attended and unattended speech sources. In this study, we investigated whether these entrainment responses can be used to measure how intelligible a speech signal is to a subject. We hypothesized that when intelligibility is degraded, attention wanes and the stimulus-response correlation will decrease. To test this, we measured a listener’s ability to detect words in noisy, natural speech while recording brain activity using EEG. We altered intelligibility by presenting congruent or incongruent video of the speaker along with their speech. For almost all subjects, word detection performance improved in the congruent condition and this improvement coincided with an increase in stimulus-response correlation. We conclude that simultaneous recordings of perceived sound and EEG activity may represent a practical tool to assess speech intelligibility, specifically in the context of hearing aid devices.
{"title":"Using EEG to Predict Speech Intelligibility","authors":"Ivan Iotzov, L. Parra","doi":"10.32470/ccn.2019.1315-0","DOIUrl":"https://doi.org/10.32470/ccn.2019.1315-0","url":null,"abstract":"Speech signals have the ability to entrain brain activity to the rapid fluctuations found in speech sounds. This entrainment can be measured using electroencephalographic (EEG) recordings and is strong enough to allow discrimination between attended and unattended speech sources. In this study, we investigated whether these entrainment responses can be used to measure how intelligible a speech signal is to a subject. We hypothesized that when intelligibility is degraded, attention wanes and the stimulus-response correlation will decrease. To test this, we measured a listener’s ability to detect words in noisy, natural speech while recording brain activity using EEG. We altered intelligibility by presenting congruent or incongruent video of the speaker along with their speech. For almost all subjects, word detection performance improved in the congruent condition and this improvement coincided with an increase in stimulus-response correlation. We conclude that simultaneous recordings of perceived sound and EEG activity may represent a practical tool to assess speech intelligibility, specifically in the context of hearing aid devices.","PeriodicalId":281121,"journal":{"name":"2019 Conference on Cognitive Computational Neuroscience","volume":"20 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1900-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133195850","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 1900-01-01DOI: 10.32470/ccn.2019.1243-0
Isabella Pozzi, S. Bohté, P. Roelfsema
{"title":"Q-AGREL: Biologically Plausible Attention Gated Deep Reinforcement Learning","authors":"Isabella Pozzi, S. Bohté, P. Roelfsema","doi":"10.32470/ccn.2019.1243-0","DOIUrl":"https://doi.org/10.32470/ccn.2019.1243-0","url":null,"abstract":"","PeriodicalId":281121,"journal":{"name":"2019 Conference on Cognitive Computational Neuroscience","volume":"60 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1900-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134600905","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 1900-01-01DOI: 10.32470/ccn.2019.1383-0
Evelyne K. Tantry, Joshua Ortiz-Guzman, B. Arenkiel
Neural circuits governing food intake have been widely studied. However, our current understanding hinges on a binary hypothalamic neuronal model that fails to address more adaptive feeding behaviors underpinning variable environmental conditions. Previous work in our lab posits an extra-hypothalamic circuit involving the cholinergic-rich diagonal band of Broca (DBB) and the valence encoding basolateral amygdala (BLA). To further analyze this circuit, we use a projection defined approach to characterize the cellular composition of the BLA. We used a stereotactic frame for bilateral injections of channelrhodopsin and tdTomato containing viruses into the DBB, and the nucleus accumbens (NAc) or the lateral hypothalamic area (LHA), respectively. The latter regions were chosen because of their established involvement in feeding. We then determined projection profiles of BLA cells using channelrhodopsin assisted circuit mapping (CRACM) and optogenetics, and found that neurons projecting to the LHA exclusively possess fast-acting nicotinic synapses, whereas neurons expressing slowacting muscarinic synapses project exclusively to the NAc. The contrasting nature these receptors indicate there to be more dynamic neural regions involved in orchestrating complex feeding behaviors.
{"title":"The Impact of Acetylcholine on Basolateral Amygdala Macrocircuits","authors":"Evelyne K. Tantry, Joshua Ortiz-Guzman, B. Arenkiel","doi":"10.32470/ccn.2019.1383-0","DOIUrl":"https://doi.org/10.32470/ccn.2019.1383-0","url":null,"abstract":"Neural circuits governing food intake have been widely studied. However, our current understanding hinges on a binary hypothalamic neuronal model that fails to address more adaptive feeding behaviors underpinning variable environmental conditions. Previous work in our lab posits an extra-hypothalamic circuit involving the cholinergic-rich diagonal band of Broca (DBB) and the valence encoding basolateral amygdala (BLA). To further analyze this circuit, we use a projection defined approach to characterize the cellular composition of the BLA. We used a stereotactic frame for bilateral injections of channelrhodopsin and tdTomato containing viruses into the DBB, and the nucleus accumbens (NAc) or the lateral hypothalamic area (LHA), respectively. The latter regions were chosen because of their established involvement in feeding. We then determined projection profiles of BLA cells using channelrhodopsin assisted circuit mapping (CRACM) and optogenetics, and found that neurons projecting to the LHA exclusively possess fast-acting nicotinic synapses, whereas neurons expressing slowacting muscarinic synapses project exclusively to the NAc. The contrasting nature these receptors indicate there to be more dynamic neural regions involved in orchestrating complex feeding behaviors.","PeriodicalId":281121,"journal":{"name":"2019 Conference on Cognitive Computational Neuroscience","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1900-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130376130","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 1900-01-01DOI: 10.32470/ccn.2019.1205-0
M. M. Nejad, Daniel Trpevski, J. Kotaleski, R. Schmidt
Striatal projection neurons (SPNs) in the basal ganglia gradually increase their firing rate during movement initiation. Arkypallidal neurons in globus pallidus briefly increase their firing rate upon a Stop signal, which cues movement cancellation. This increase potentially leads to the suppression of movement-related activity in striatum by inhibiting SPNs. However, this brief inhibition from arkypallidal neurons may be too short to completely prevent the gradual firing rate increase in SPNs. Here, we investigated the impact of the brief inhibition on the gradual increase in a multi-compartmental model of a SPN. We reproduced the movement-related firing pattern in the SPN model neuron by brief clustered excitation added to a baseline, subthreshold excitation. This brief clustered excitation evoked a dendritic plateau potential leading to a long-lasting depolarization at the soma, which enhanced the somatic excitability and evoked spikes upon the baseline excitation that was formerly subthreshold. A brief inhibition, representing arkypallidal stop responses, applied on the dendritic site where the clustered excitation was present, suppressed the somatic depolarization and attenuated the movement-related activity similar to the firing pattern observed in rats for successful action suppression. We conclude that arkypallidal Stop responses can suppress movement-related activity in the striatum by suppressing the dendritic plateau potentials.
{"title":"Stopping actions by suppressing striatal plateau potentials","authors":"M. M. Nejad, Daniel Trpevski, J. Kotaleski, R. Schmidt","doi":"10.32470/ccn.2019.1205-0","DOIUrl":"https://doi.org/10.32470/ccn.2019.1205-0","url":null,"abstract":"Striatal projection neurons (SPNs) in the basal ganglia gradually increase their firing rate during movement initiation. Arkypallidal neurons in globus pallidus briefly increase their firing rate upon a Stop signal, which cues movement cancellation. This increase potentially leads to the suppression of movement-related activity in striatum by inhibiting SPNs. However, this brief inhibition from arkypallidal neurons may be too short to completely prevent the gradual firing rate increase in SPNs. Here, we investigated the impact of the brief inhibition on the gradual increase in a multi-compartmental model of a SPN. We reproduced the movement-related firing pattern in the SPN model neuron by brief clustered excitation added to a baseline, subthreshold excitation. This brief clustered excitation evoked a dendritic plateau potential leading to a long-lasting depolarization at the soma, which enhanced the somatic excitability and evoked spikes upon the baseline excitation that was formerly subthreshold. A brief inhibition, representing arkypallidal stop responses, applied on the dendritic site where the clustered excitation was present, suppressed the somatic depolarization and attenuated the movement-related activity similar to the firing pattern observed in rats for successful action suppression. We conclude that arkypallidal Stop responses can suppress movement-related activity in the striatum by suppressing the dendritic plateau potentials.","PeriodicalId":281121,"journal":{"name":"2019 Conference on Cognitive Computational Neuroscience","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1900-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125698441","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 1900-01-01DOI: 10.32470/ccn.2019.1115-0
D. G. Nagy, B. Török, Gergő Orbán
From a continuous stream of experience, how does the human brain determine what parts to remember and what to forget? It is extensively documented that humans are prone to systematic biases in these decisions. Such systematic biases are often construed as byproducts of adaptive processes. We argue that the computational resource constraints on memory can be formalised in the normative framework of lossy compression and that optimal adaptation to the constraints can be achieved by exploiting a generative model of the environment for compression. Recent advances in machine learning yielded powerful tools to approximate such solutions. In this study, we harness these advances to show that generative compression can explain a wide variety of memory phenomena including the effects of domain expertise on recall, gist based distortions in recalling lists of semantically related words and the influence of contextual cues in memory for hand drawn sketches.
{"title":"Rate distortion trade-off in human memory","authors":"D. G. Nagy, B. Török, Gergő Orbán","doi":"10.32470/ccn.2019.1115-0","DOIUrl":"https://doi.org/10.32470/ccn.2019.1115-0","url":null,"abstract":"From a continuous stream of experience, how does the human brain determine what parts to remember and what to forget? It is extensively documented that humans are prone to systematic biases in these decisions. Such systematic biases are often construed as byproducts of adaptive processes. We argue that the computational resource constraints on memory can be formalised in the normative framework of lossy compression and that optimal adaptation to the constraints can be achieved by exploiting a generative model of the environment for compression. Recent advances in machine learning yielded powerful tools to approximate such solutions. In this study, we harness these advances to show that generative compression can explain a wide variety of memory phenomena including the effects of domain expertise on recall, gist based distortions in recalling lists of semantically related words and the influence of contextual cues in memory for hand drawn sketches.","PeriodicalId":281121,"journal":{"name":"2019 Conference on Cognitive Computational Neuroscience","volume":"12 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1900-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132441996","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：