Figure - available from: PLOS Computational Biology
This content is subject to copyright.
Hierarchical network regime III winner-take-all operation increases offer separability (A) Example offer where Regime III improves in relative performance as uncertainty increases. (B) Linear and max operations used to calculate decision value distributions for the offer shown in (A). In a linear operation, the mean and variance of both attribute are used to calculate distribution from which the final offer value is drawn. The max operation first selects the larger within-attribute alternative, which then becomes the decision value distribution. (C) Decision value distributions produced by the max operation maintain greater separability as uncertainty increases. Distributions are shown for the decision values calculated in (B) using σ η I 2 = 0 . 4 (top row), σ η I 2 = 0 . 75 (middle row) and σ η I 2 = 1 . 5 (bottom row). In the left column are results for the linear operation, and in the right from the max operation. (D) For those offers where regime III performance (J+ = 0.35, J− = −0.1) exceeds that of the LN, a max operation provides greater separability between decision value distributions. For offers that transitioned to better HN performance, at each uncertainty level we calculated the d′ of decision value distributions produced by the linear operation and the max operation. Shown are the mean and standard error of the d′s.
Source publication
We are constantly faced with decisions between alternatives defined by multiple attributes, necessitating an evaluation and integration of different information sources. Time-varying signals in multiple brain areas are implicated in decision-making; but we lack a rigorous biophysical description of how basic circuit properties, such as excitatory-i...
Citations
... Under the notion that uncertainty requires exploration of a larger space of options, we argue that this is akin to a lower learning rate for an individual feature at the benefit of distributed learning across uncertain features. Non-selective gain increases, e.g., provided by global arousal, can favor such distributed learning 108 . We observe that pupil sensitivity to rising uncertainty is retained across the adult lifespan but dampens in older age. ...
The ability to prioritize among input features according to relevance enables adaptive behaviors across the human lifespan. However, relevance often remains ambiguous, and such uncertainty increases demands for dynamic control. While both cognitive stability and flexibility decline during healthy ageing, it is unknown whether aging alters how uncertainty impacts perception and decision-making, and if so, via which neural mechanisms. Here, we assess uncertainty adjustment across the adult lifespan (N = 100; cross-sectional) via behavioral modeling and a theoretically informed set of EEG-, fMRI-, and pupil-based signatures. On the group level, older adults show a broad dampening of uncertainty adjustment relative to younger adults. At the individual level, older individuals whose modulation more closely resembled that of younger adults also exhibit better maintenance of cognitive control. Our results highlight neural mechanisms whose maintenance plausibly enables flexible task-set, perception, and decision computations across the adult lifespan.
... The brain is comprised of billions of neurons, which are interconnected through synapses, enabling the execution of intricate activities and tasks [1][2][3][4][5][6]. Many network models have been proposed to explain the complex functions of the brain, such as star-coupled networks, random networks, small-world networks, and scale-free networks [7][8][9][10]. ...
Transmission of weak signals in neural networks is crucial for understanding the functionality of brain. In this work, stochastic resonance (SR) in the three neuron FitzHugh–Nagumo (FHN) motifs and its small-world network with higher order motif interactions are studied. Simulation results show that a single motif induces SR and responds better to high-frequency weak signal. Stronger coupling strength within the motif increases the firing rate of the output neurons, resulting in a more pronounced resonance. Considering only the connections within the motif, a higher in-degree of the output neuron or a shorter minimum path length between input and output neurons will lead to a better response to weak signals. SR phenomena can also be observed in small-world networks composed of these motif. Increasing whether the motif coupling or node coupling strength enhances the firing rate of output neurons, amplifying the response. There is a very strong correlation between firing rate of output neurons and response. Our results may provide insights into the propagation of weak signals in higher order networks and the selection of appropriate network topology.
... A fundamental understanding of the complex interac ons between excita on and inhibi on is thought to be cri cal to the development of effec ve therapeu cs (9). Accordingly, numerous models of network func oning and experimental studies have explored the consequences of altering the E:I balance through systema c manipula on of either the excita on or the inhibi on component (7,(45)(46)(47)(48)(49)(50)(51)(52)(53). For example, many models of a ractor networks posit that inhibi on throughout the network remains stable, serving as a global check, while excitatory drive is more dynamic and acts as the driving factor in varia ons to the overall E:I balance. ...
... Examining the excitatory and inhibitory coupling in our data independently, we found that inhibitory tone was more ghtly linked to ongoing LFP dynamics than excitatory tone (Fig 4A- 37). This is in direct contrast to most computa onal models which suggest that changes in excitatory ac vity drive changes in E:I balance (46,48,49), but more in line with dynamic changes in a network's inhibitory tone in response to varying levels of environmental uncertainty (47). ...
The balance between excitation and inhibition is critical to brain functioning, and dysregulation of this balance is a hallmark of numerous psychiatric conditions. Measuring this excitation-inhibition (E:I) balance in vivo has remained difficult, but theoretical models have proposed that characteristics of local field potentials (LFP) may provide an accurate proxy. To establish a conclusive link between LFP and E:I balance, we recorded single units and LFP from the prefrontal cortex (mPFC) of rats during decision making. Dynamic measures of synaptic coupling strength facilitated direct quantification of E:I balance and revealed a strong inverse relationship to broadband spectral power of LFP. These results provide a critical link between LFP and underlying network properties, opening the door for non-invasive recordings to measure E:I balance in clinical settings.
... In other words, the decision weight with a larger magnitude could be detected earlier in time, appearing as the time asynchrony we observed in the ESTP distributions. To address this concern, we applied a neural circuit model of multiattribute decision-making (41) to elucidate the impacts of magnitude and temporal asynchrony on the dynamics of the decision weights. ...
Two behavioral motivations coexist in transgressors following an interpersonal transgression—approaching and compensating the victim and avoiding the victim. Little is known about how these motivations arise, compete, and drive transgressors’ decisions. The present study adopted a social interaction task to manipulate participants’ (i.e., the transgressor) responsibility for another’s (i.e., the victim) monetary loss and measure the participants’ tradeoff between compensating the victim and avoiding face-to-face interactions with the victim. Following each transgression, participants used a computer mouse to choose between two options differing in the amount of compensation to the victim and the probability of face-to-face contact with the victim. Results showed that as participants’ responsibility increased, 1) the decision weights on contact avoidance relative to compensation increased, and 2) the onset of the contact-avoidance attribute was expedited and that of the compensation attribute was delayed. These results demonstrate how competing social motivations following transgression evolve and determine social decision-making and shed light on how social-affective state modulates the dynamics of decision-making in general.
... Such selective processing may be akin to sitting deep in a single attractor within a landscape of potential processing modes, a phenomenon linked to cortical alpha rhythms 155 that likely emerge from relatively selective thalamocortical loop interactions 156 . Conversely, when we are less certain about what to attend to in our environment, the brain may instead need to track multiple stimulus features at the same time 157,158 . Doing so may require a 'flatter' attractor landscape, within which the brain can more easily switch processing modes, a process likely requiring higher neural 'excitability' (that is, variability) and neuromodulatory arousal 159 via more distributed interactions between the thalamus and the cerebral cortex 14,62 . ...
The thalamus is a small, bilateral structure in the diencephalon that integrates signals from many areas of the CNS. This critical anatomical position allows the thalamus to influence whole-brain activity and adaptive behaviour. However, traditional research paradigms have struggled to attribute specific functions to the thalamus, and it has remained understudied in the human neuroimaging literature. Recent advances in analytical techniques and increased accessibility to large, high-quality data sets have brought forth a series of studies and findings that (re-)establish the thalamus as a core region of interest in human cognitive neuroscience, a field that otherwise remains cortico-centric. In this Perspective, we argue that using whole-brain neuroimaging approaches to investigate the thalamus and its interaction with the rest of the brain is key for understanding systems-level control of information processing. To this end, we highlight the role of the thalamus in shaping a range of functional signatures, including evoked activity, interregional connectivity, network topology and neuronal variability, both at rest and during the performance of cognitive tasks.
... To test how learned predictive reward values could drive attention, we compared three possible relationships between these values and attention (in addition to no relationship): summation, absolute difference, and maximum (see Materials and Methods). All of these functions could be implemented by canonical recurrent neural network circuits and had been used in prior studies to calculate attention based on subjective reward values, sometimes in much simpler reward schedules where these possibilities could be hard to distinguish from each other (Anderson et al., 2011;Hunt et al., 2014;Niv et al., 2015;Soltani et al., 2016Soltani et al., , 2021Leong et al., 2017;Farashahi et al., 2017b;Gluth et al., 2018;Daniel et al., 2020;Farashahi and Soltani, 2021;Pettine et al., 2021). In our model, predictive reward values of the two presented options along different dimensions are first passed through one of the above functions and then normalized across dimensions to have a sum of 1, resulting in one attention weight per feature or conjunction dimension (e.g., color or color/shape conjunction). ...
Real-world choice options have many features or attributes, whereas the reward outcome from those options only depends on a few features/attributes. It has been shown that humans learn and combine feature-based with more complex conjunction-based learning to tackle challenges of learning in complex reward environments. However, it is unclear how different learning strategies interact to determine what features should be attended and control choice behavior, and how ensuing attention modulates future learning and/or choice. To address these questions, we examined human behavior during a three-dimensional learning task in which reward outcomes for different stimuli could be predicted based on a combination of an informative feature and conjunction. Using multiple approaches, we first confirmed that choice behavior and reward probabilities estimated by participants were best described by a model that learned the predictive values of both the informative feature and the informative conjunction. In this model, attention was controlled by the difference in these values in a cooperative manner such that attention depended on the integrated feature and conjunction values, and the resulting attention weights modulated learning by increasing the learning rate on attended features and conjunctions. However, there was little effect of attention on decision making. These results suggest that in multidimensional environments, humans direct their attention not only to selectively process reward-predictive attributes, but also to find parsimonious representations of the reward contingencies for more efficient learning.
Significance Statement
From trying exotic recipes to befriending new social groups, outcomes of real-life actions depend on many factors, but how do we learn the predictive values of those factors based on feedback we receive? It has been shown that humans simplify this problem by focusing on individual factors that are most predictive of the outcomes but can extend their learning strategy to include combinations of factors when necessary. Here, we examined interaction between attention and learning in a multidimensional reward environment that requires learning about individual features and their conjunctions. Using multiple approaches, we found that learning about features and conjunctions control attention in a cooperative manner and that the ensuing attention mainly modulates future learning and not decision making.
... We first show that, although the dominant low-rank structure is set on average by the mean synaptic weights [40][41][42], a perturbative approach accurately accounts for the components of individual neurons on the dominant eigenvectors arising from individual instances of the random connectivity. As a result, our low-rank approximation analytically captures the activity of individual neurons in the original E-I network defined based on local statistics. ...
... Reciprocal correlations between z ij and z ji directly lead to correlations between Z and Z > and therefore a non-zero covariance σ nm between elements of m and n, that can be predicted by mean field theory ( Fig 3E, Eq (72)). The strength of the covariance between eigenvector entries reflects the strength of the additional feedback loop due to reciprocal correlations, and is therefore directly related to the deviations of the outlying eigenvalue from the uncorrelated value λ 0 (Eqs (42) and (96), Fig 3F). When the network has both homogeneous variance parameters and correlation parameters, the excitatory and inhibitory popula- (97)). ...
How the connectivity of cortical networks determines the neural dynamics and the resulting computations is one of the key questions in neuroscience. Previous works have pursued two complementary approaches to quantify the structure in connectivity. One approach starts from the perspective of biological experiments where only the local statistics of connectivity motifs between small groups of neurons are accessible. Another approach is based instead on the perspective of artificial neural networks where the global connectivity matrix is known, and in particular its low-rank structure can be used to determine the resulting low-dimensional dynamics. A direct relationship between these two approaches is however currently missing. Specifically, it remains to be clarified how local connectivity statistics and the global low-rank connectivity structure are inter-related and shape the low-dimensional activity. To bridge this gap, here we develop a method for mapping local connectivity statistics onto an approximate global low-rank structure. Our method rests on approximating the global connectivity matrix using dominant eigenvectors, which we compute using perturbation theory for random matrices. We demonstrate that multi-population networks defined from local connectivity statistics for which the central limit theorem holds can be approximated by low-rank connectivity with Gaussian-mixture statistics. We specifically apply this method to excitatory-inhibitory networks with reciprocal motifs, and show that it yields reliable predictions for both the low-dimensional dynamics, and statistics of population activity. Importantly, it analytically accounts for the activity heterogeneity of individual neurons in specific realizations of local connectivity. Altogether, our approach allows us to disentangle the effects of mean connectivity and reciprocal motifs on the global recurrent feedback, and provides an intuitive picture of how local connectivity shapes global network dynamics.
... We obtain a rank-R approximation J R of J by keeping the first R terms in Eq. (42): ...
How the connectivity of cortical networks determines the neural dynamics and the resulting computations is one of the key questions in neuroscience. Previous works have pursued two complementary strategies to quantify the structure in connectivity, by specifying either the local statistics of connectivity motifs between small groups of neurons, or by defining network-wide low-rank patterns of connectivity that determine the resulting low-dimensional dynamics. A direct relationship between these two approaches is however currently missing,
and in particular it remains to be clarified how local connectivity statistics are related to the global connectivity structure and shape the low-dimensional activity. To bridge this gap, here we develop a method for mapping local connectivity statistics onto an approximate global low-rank structure. Our method rests on approximating the global connectivity matrix using dominant eigenvectors, which we compute using perturbation theory for random matrices. This approach demonstrates that multi-population networks defined from local connectivity properties can in general be approximated by low-rank connectivity with Gaussian-mixture statistics. We specifically apply this method to excitatory-inhibitory networks, and show that it leads to accurate predictions for both the low-dimensional dynamics, and for the activity of individual neurons. Altogether, our approach allows us to disentangle the effects of mean connectivity and reciprocal motifs on the global recurrent feedback, and provides an intuitive picture of how local connectivity shapes global network dynamics.
... Notably, neuronal recordings in monkeys performing the same choice task used here indicated that decisions rely on circuit inhibition 26,40 . Furthermore, current computational models suggest that value comparison relies on a balance between recurrent excitation and pooled inhibition 6,41,42 . In these models, increasing inhibition makes decisions less accurate 43,44 . ...
Economic choices between goods entail the computation and comparison of subjective values. Previous studies examined neuronal activity in the orbitofrontal cortex (OFC) of monkeys choosing between different types of juices. Three groups of neurons were identified: offer value cells encoding the value of individual offers, chosen juice cells encoding the identity of the chosen juice, and chosen value cells encoding the value of the chosen offer. The encoded variables capture both the input (offer value) and the output (chosen juice, chosen value) of the decision process, suggesting that values are compared within OFC. Recent work demonstrates that choices are causally linked to the activity of offer value cells. Conversely, the hypothesis that OFC contributes to value comparison has not been confirmed. Here we show that weak electrical stimulation of OFC specifically disrupts value comparison without altering offer values. This result implies that neuronal populations in OFC participate in value comparison.
... In sum, pupil-linked arousal may provide a window into yet another cortical circuit property, cortical E/I, relevant for cognitive computation (Cavanagh et al., 2020;Kosciessa et al., 2021;Lam et al., 2017;Murphy and Miller, 2003;Pettine et al., 2021;Pfeffer et al., 2021). ...
Fluctuations in arousal, controlled by subcortical neuromodulatory systems, continuously shape cortical state, with profound consequences for information processing. Yet, how arousal signals influence cortical population activity in detail has so far only been characterized for a few selected brain regions. Traditional accounts conceptualize arousal as a homogeneous modulator of neural population activity across the cerebral cortex. Recent insights, however, point to a higher specificity of arousal effects on different components of neural activity and across cortical regions. Here, we provide a comprehensive account of the relationships between fluctuations in arousal and neuronal population activity across the human brain. Exploiting the established link between pupil size and central arousal systems, we performed concurrent magnetoencephalographic (MEG) and pupillographic recordings in a large number of participants, pooled across three laboratories. We found a cascade of effects relative to the peak timing of spontaneous pupil dilations: Decreases in low-frequency (2-8 Hz) activity in temporal and lateral frontal cortex, followed by increased high-frequency (>64 Hz) activity in mid-frontal regions, followed by monotonic and inverted-U relationships with intermediate frequency-range activity (8-32 Hz) in occipito-parietal regions. Pupil-linked arousal also coincided with widespread changes in the structure of the aperiodic component of cortical population activity, indicative of changes in the excitation-inhibition balance in underlying microcircuits. Our results provide a novel basis for studying the arousal modulation of cognitive computations in cortical circuits.
