As we go about our days, how often do we fidget, compared to how frequently we make larger movements, like walking down the hall? And how rare is a trek across town compared to that same walk down the hall? Animals tend to follow a mathematical law that relates the size of our movements to how often we do them.

This law posits that small-to-medium movements and large-to-huge movements are related in the same way, that is, the law is ‘scale-free’, it holds the same for different scales of movement. Surprisingly, measurements of brain activity also follow this scale-free law: the level of activation of a group of neurons relates to how often they are activated in the same way for different levels of activation.

Although body movements and brain activity behave in a mathematically similar way, these two facts had not previously been linked. Jones et al. studied body movements and brain activity in mice, and found that scale-free body movements were linked to scale-free brain activity, but only in certain subsets of neurons. This observation had been hidden because other subsets of neurons compete with scale-free neurons. When the scale-free neurons turn on, the competing groups turn off. When averaged together, these fluctuations cancel out.

The findings of Jones et al. provide a new understanding of how brain and body dynamics are orchestrated in healthy organisms. In particular, their results suggest that the complex, multi-scale nature of behavior and body movements may emerge from brain activity operating at a critical tipping point between order and disorder, at the edge of chaos.

From fidgeting to expeditions, natural body movements manifest over a very broad range of spatiotemporal scales. The complexity of such movements is often organized with fractal structure, scale-free fluctuations spanning multiple spatiotemporal orders of magnitude (Anteneodo and Chialvo, 2009; Proekt et al., 2012; Sims et al., 2008; Viswanathan et al., 1999). Such behavioral complexity may be beneficial for foraging (Garg and Kello, 2021; Sims et al., 2008; Viswanathan et al., 1999; Wosniack et al., 2017), visual search (Viswanathan et al., 1999), decision-making based on priority (Barabási, 2005; Sorribes et al., 2011), flexible switching of behavior (Abe, 2020), and perhaps more. Similarly, fluctuations of ongoing neural activity in the cerebral cortex can exhibit fractal, scale-free fluctuations like the spatiotemporal cascades sometimes referred to as ‘neuronal avalanches’ (Beggs and Plenz, 2003; Bellay et al., 2015; Ma et al., 2019; Priesemann et al., 2014; Scott et al., 2014; Shew et al., 2015; Shriki et al., 2013; Tagliazucchi et al., 2012; Yu et al., 2017) and long-range temporal correlations (Hardstone et al., 2012; Kello et al., 2010; Palva et al., 2013; Smit et al., 2013). Considering these observations of behavior and brain activity together, a simple, yet unanswered question arises. Are scale-free dynamics in cerebral cortex related to scale-free body movements? Or are these statistical similarities merely superficial, without any direct link?

Two previous studies in humans have shown that temporally scale-free ongoing brain dynamics are related to some subtle aspects of behavior – fluctuations in success at rhythmic finger tapping (Smit et al., 2013) and sensory detection tasks (Palva et al., 2013). But, here, we are concerned with less subtle, larger-scale body movements and invasive measurements of brain activity with single neuron resolution.

According to prevailing theories of scale-free ongoing neural activity, these dynamics are often interpreted as ‘background’ activity, not directly linked to behavior. This view arises in part because computational models of scale-free neural activity generate this type of dynamics autonomously, due to internal interactions, without explicit modeling of a body that behaves (Dalla Porta and Copelli, 2019; di Santo et al., 2018; Girardi-Schappo et al., 2020; Li and Shew, 2020; Wilting et al., 2018). Similarly, in experiments, scale-free neural activity has typically been observed in animals that are at rest, not engaged by any particular motor task or sensory input, or under anesthesia (Fontenele et al., 2019; Gautam et al., 2015; Gireesh and Plenz, 2008), or in vitro (Beggs and Plenz, 2003; Bellay et al., 2020; Shew et al., 2009; Tetzlaff et al., 2010) where there is no behavior to be observed. Based on these previous experimental observations and modeling efforts, one might naturally conclude that scale-free neural activity is internally generated, emerging spontaneously, without any link to behavior.

However, recent experiments in awake mice have revealed that behaviors typically ignored by experimenters, such as whisking, eye and face movement, fidgeting, walking, and running, are strongly related to ongoing neural activity in the cortex (Clancy et al., 2019; Musall et al., 2019; Salkoff et al., 2020; Stringer et al., 2019). These studies did not address our question here about scale-free-ness of neural activity and behavior, but they suggest that if ongoing activity is scale-free, then it may be strongly related to behavior. More specifically, these studies point out the possibility that previous observations of scale-free neural activity in awake animals didn’t notice a link to behavior because those studies did not measure ongoing, spontaneous movements of the face, whiskers, and body. If this is the case, it would dramatically revise the typical interpretation of scale-free ongoing activity and provide direct evidence that scale-free behavior and scale-free brain activity are inseparably related. However, in the work by Stringer et al., 2019, the measured behaviors could explain only 10–30% of the total variance in neural activity fluctuations. Thus, the question remains: is this behaviorally-related 10–30% of neural activity scale-free? Here, our aim was to answer this question.

Here, we analyzed simultaneous recordings of more than 10,000 pyramidal neurons during 1–2 hr in visual cortex of awake mice ( seven mice, nine recordings, recording duration 1.52 ± 0.51 hr), performed and first reported by Stringer et al., 2019. The neurons were distributed throughout a volume (300–400 μm in-depth, 0.9 × 0.935 mm in the area) and sampled at 3 Hz. In addition to the neural activity, several aspects of behavior and body movements were also recorded (Figure 1). Here, we focused on four aspects of behavior. First, we studied run speed, which was assessed using an optical sensor and a floating spherical treadmill. Second, we examined pupil diameter, which was obtained from a camera targeting the faces of the mice. Pupil diameter dynamics are associated with changes in arousal and other aspects of body movement and sensory input (Reimer et al., 2016; Reimer et al., 2014; Vinck et al., 2015). Third, we studied changes in direction of gaze by tracking the speed of the center of the pupil as the mice looked around. Finally, we used previously developed methods to study whisker motion (Stringer et al., 2019), which was also obtained from the face camera.

Figure 1

Download asset Open asset

Scale-free behavior

The behavior tended to occur in bursts; rest periods were punctuated with well-defined bouts of body movement (Figure 1C and Figure 2A). All four of the behavioral variables we studied tended to start and stop together, approximately, but differed in their detailed, fast fluctuations (Figure 1C). We first sought to determine whether the behaviors exhibited scale-free dynamics. To this end, we first defined each bout of elevated body activity, hereafter called a behavioral ‘event,’ based on a threshold (the median of the entire time series). Each behavioral event began when the behavioral time series, run speed for example (Figure 2A), exceeded the threshold and ended when it returned below the threshold. The ‘size’ of each behavioral event was defined as the area between the threshold and the variable during the event (Figure 2A). This definition of events and their sizes is motivated by previous studies of scale-free neural activity (Gautam et al., 2015; Larremore et al., 2014; Li and Shew, 2020; Poil et al., 2012), typically referred to as neuronal avalanches. In the case of run speed, the event size corresponds to an effective distance traveled during the event. If the animal behavior is scale-free, one expects the distribution Pr(s) of behavioral event sizes (s) to have a power-law form, Pr(s)~s^-τ. We found that these distributions often were in the form of a power-law over a wide range of event sizes (Figure 2B). For example, the run speed event size distribution for mouse 1 was scale-free for nearly four orders of magnitude of event sizes (power-law range, r=3.9 decades). To rigorously assess the range over which the data is power-law distributed, we used a maximum likelihood fitting algorithm that accounts for the number of events observed and the possibility of a cutoff at the head and tail of the distribution (Methods). Our approach builds on that used in previous studies (Shew et al., 2015). We report the power-law range r in decades, i.e., the number of orders of magnitude. If no range meets our fit criteria for statistical significance, we report r=0.

Figure 2 with 2 supplements see all

Download asset Open asset

Although not all behaviors were scale-free over such a large range for all recordings, we found that most of the mice exhibited scale-free behaviors, particularly for running and pupil fluctuations (Figure 2C). In addition to the power-law range for larger behavioral events, some behaviors exhibited a range of small-scale behavioral events with a flat distribution consistent with low amplitude noise in the measurement methods (Figure 2—figure supplement 1). These conclusions were robust to changes in the two most important parameters for the power-law fitting algorithm: the goodness-of-fit criterion and the outlier exclusion threshold (more on outlier exclusion in Figure 2—figure supplement 2).

Scale-free neural activity

Next, we turned to the neural activity to test whether it was scale-free like the behavior. First, we averaged over all neurons to obtain a single population activity time series. Then, we treated the neural data in a similar way to the behavioral data. We defined neural events with a median threshold and defined event size as the area between the threshold and the data during the event (Figure 3A). This definition of events and event sizes is well-established in previous studies of scale-free neural activity (Gautam et al., 2015; Larremore et al., 2014; Li and Shew, 2020; Poil et al., 2012). (For more discussion of why this definition was chosen over other established alternatives, see Methods). The resulting distributions of neural event sizes were not scale-free; they were poorly fit by a power-law distribution (Figure 3B). Considering all 9 recordings, we found that r=1.5 ± 0.9 decades (mean ± SD). Such a short power-law range should be interpreted as evidence against any power-law at all, because nearly any function can be fit on a short range of data (more on why a large power-law range is important below).

Figure 3 with 2 supplements see all

Download asset Open asset

One note of warning: at first glance, the distribution in Figure 3B may appear to have a decent power-law fit for the small event sizes. This is a misleading artifact resulting from using logarithmic bins and very few samples to estimate probability density. The inset in Figure 3B shows the number of events rather than probability density. See also Figure 2—figure supplement 1 for more clarification of this point. We emphasize that our power-law fitting algorithm does not use any bins and accounts for sample size, thus avoiding such misleading artifacts.

The lack of scale-free dynamics for the neural population was surprising considering two facts together: first, the behavior exhibited scale-free dynamics (Figure 2), and second, many neurons are strongly correlated with behavior (Figure 1 and Stringer et al., 2019). One possible explanation for this puzzling observation is that summing the entire population together obscures the scale-free dynamics of certain subsets of neurons. Next, we set out to test this possibility. Since we did not know, a priori, which neurons might be in these subsets, we adopted a brute-force, shotgun search. First, we picked a ‘seed’ neuron at random from the entire population. Then we identified the 50 neurons that were most correlated with the seed neuron (we also tried 200 neurons without major impacts on the following results, Figure 3E). We averaged the activity of these 50 neurons to obtain a single time series and proceeded to define events and examine their size distribution. Two examples of time series obtained from subsets of neurons in this way are shown in Figure 3A. We repeated this process for 1000 different seed neurons. We found that some subsets of neurons were indeed scale-free, with power-law scaling over more than four orders of magnitude, while other subsets were not scale-free (Figure 3C). With so many neurons to choose from and a shotgun approach like this, it is important to avoid chance-level false positive conclusions. As a conservative control for this possibility, we repeated the analysis, but with surrogate data, generated by applying a random time shift to each neuron’s activity time series relative to other neurons. This time-shifted control data, thus, has identical statistics to the original data at the single neuron level, but correlations among neurons occur only by chance. For time-shifted controls, we found that large power-law ranges (greater than 3.5 decades) were rare and not found at all in the majority of recordings (Figure 3D and E). This conclusion was robust for different choices of group size and goodness-of-fit criteria used for the power-law fitting (Figure 3E).

Two concerns with the data analyzed here are that it was sampled at a slow time scale (3 Hz frame rate) and that the deconvolution methods used to obtain the data here from the raw GCAMP6s Ca imaging signals are likely to miss some of the underlying spike activity (Huang et al., 2021). Since our analysis of neural events hinges on summing up activity across neurons, could it be that the missed activity creates systematic biases in our observed event size statistics? To address this question, we analyzed some time-resolved spike data (Neuropixel recording from Stringer et al., 2019). Starting from the spike data, we created a slow signal, similar to that we analyzed here by convolving with a Ca-transient, down sampling, deconvolving, and z-scoring (Figure 3—figure supplements 1–2). We compared neural event size distributions to ‘ground truth’ based on the original spike data (with no loss of spikes) and found that the neural event size distributions were very similar, with the same exponent and the same power-law range (Figure 3—figure supplement 1). Thus, we conclude that our reported neural event size distributions are reliable.

Linking scale-free brain activity and behavior

Having established that certain subsets of neurons have scale-free dynamics and that behavior has scale-free dynamics, we next sought to assess how the neural and behavioral dynamics are related. We first computed the correlation coefficient between each behavior time series and each neural subset time series. The correlation was very weak between the total neural population time series and behavior (red x’s in Figure 4). For the 50-neuron subsets, we found a very wide range of behavioral correlation values, from 0.9 to –0.75 (Figure 4). Interestingly, the subsets with the greatest correlation with behavior also tended to have the largest ranges of power-law scaling. Moreover, the neural subsets with the smallest power-law range were most often near zero correlation with behavior. These results strongly suggest that the scale-free subsets of neurons are directly related to scale-free behavior. It is also interesting to note that many of the subsets that were strongly anticorrelated with behavior also had a large power-law range (the U-shaped relationships in Figure 4). These findings were very robust in 6 out of 9 recordings (mouse 1–4 c, Figure 4), but less prominent for some behaviors in the remaining three recordings (mouse 5–7). We note that these three ‘outlier’ animals also had tendencies to run much more than usual (mouse 6 and 7 ran 50 times more than the next most active animal) or much less than usual (mouse 5 ran 10 times less than the next least active animal) as shown in Figure 4—figure supplement 1. This observation suggests that our main findings hold best in vigilance states that are moderate, not too hyperactive, or hypoactive.

Figure 4 with 1 supplement see all

Download asset Open asset

Power-law range, exponents, and scaling relations

Previous studies of scale-free neural activity have typically not reported power-law range. Readers may wonder why we focused our analyses here on the power-law range. So far, our results offer two answers to this question. First, and most importantly, only neural subsets with large power-law ranges were strongly correlated with behavior (Figure 4). Second, any claim of power-law scaling is stronger if that scaling extends over a greater range. Previous studies often rely on a ‘rule of thumb’ that one or two decades of scaling is not bad. For the data analyzed here, we found that a power-law range of 1–2 decades is insignificant compared to chance for event size distributions (Figure 3).

One reason previous studies of scale-free brain activity have not reported power-law range stems from their motivating hypothesis that scale-free activity results from the neural circuits operating in a special dynamical regime near the tipping point of a phase transition, i.e., near criticality. Rather than power-law range, these studies focused on scaling relationships between the power-law exponents of the event size and duration distributions as predicted by the theory of critical phenomena. Based on our results, we contend that these previous studies might be strengthened by also accounting for power-law range. Conversely, previous studies also raise a natural question about whether the data we analyzed here supports predictions from the theory of criticality. Our next goal is to answer this question.

According to prevailing theories of criticality in neural systems, both the sizes and durations of events should be power-law distributed. Moreover, the theory of crackling noise, which explains a broad class of stick-slip critical phenomena (Sethna et al., 2001), predicts that the size power-law exponent τ and the duration power-law exponent α should be related to how size scales with duration. More specifically, S(T)~T^β where β_crackling = (α - 1)/(τ –1). Many previous observations of scale-free neural activity support this prediction with values of β_crackling ≈ 1.1–1.3 (Fontenele et al., 2019; Fosque et al., 2021; Friedman et al., 2012; Ma et al., 2019; Shew et al., 2015). To test this theory, we first found the best-fit power-law and its exponent τ for the size distribution (Figure 5A). Then, we fit a power-law to the event duration distributions for the same set of events that were in the range of power-law scaling for event sizes (Figure 5B), thus obtaining α. And finally, we computed a best-fit β by doing a linear fit to log(S) versus log(T) as shown in Figure 5C. We found that τ, α, and β varied dramatically across different neural subsets (Figure 5D). However, if we examined only the behaviorally-correlated subsets, i.e., those with a large power-law range (r>3.5), we found that the power-law exponents were relatively consistent: τ=1.0 ± 0.1, α=1.2 ± 0.2, and fit β=2.2 ± 0.2 (mean ± SD, Figure 5E). Thus, we conclude that power-law scaling exists for both sizes and durations of neural events and that size scales with duration according to another power-law. The existence of these scaling laws is consistent with criticality, but we found that our measured values of β were not consistent with crackling noise theory predictions; best fit β was not close to (α - 1)/(τ –1). This was typically because τ was near one, resulting in small, sometimes negative values of τ–1 in the denominator of the predicted β formula. Moreover, our observations of β were greater than two, which is not close to previously reported values from experiments – around 1.1–1.3. This suggests that if the scale-free neural activity we observe has its origins in criticality, it may be a type of criticality that is different than previous observations. Indeed, this discrepancy with the existing theory is part of the motivation for the new model we present below. Additional scaling laws from the experiments are compared to the model results below.

Figure 5

Download asset Open asset

Finally, we examined the exponents for distributions of behavioral events. Considering the behaviors with a large power-law range (r>3), we found τ=1.2 ± 0.2, α=1.4 ± 0.3, and fit β=1.8 ± 0.4. These exponents were related to the exponents and power-law ranges for neural event distributions, but only if we considered the neural subsets with large power-law range (r>3.5 decades) and strong correlation to behavior (R>0.6). We found that τ, α, r, and β were significantly correlated for neural and behavioral events (Pearson correlation R_τ=0.5, p<10^–17; R_α=0.2, p<0.002; R_r = 0.4, p<10^–66; R_β=0.3, p<10^–4). For these comparisons, we paired each neural subset with the type of behavior that was most correlated to that subset.

Correlations of individual behavioral and neural events

The calculation of correlations between neural subsets and behavior, like those in Figure 4, represents the entire recording duration, leaving open several questions. Are the high correlations in Figure 4 simply due to coarse, on-off dynamics of the neural and behavioral activity, or are there strong correlations at the more detailed level of the faster fluctuations during a single behavioral event? To answer these questions, we computed event-specific correlations – computed between the behavior time series and the neural subset time series during each behavioral event, for each neural subset (Figure 6A). For instance, for 1000 neural subsets and 1000 behavioral events (which are typical numbers for this data), we would calculate 1 million correlation values. Each correlation is based on the relatively short time series between the start and stop of the corresponding behavioral event. These event-specific correlations were often quite high (>0.9). However, the chance-level occurrence of such high correlations can also be high for such short time series. We account for such chance-level correlations by repeating the calculations for time-shifted control data (Methods), which defines the event-specific, subset-specific chance-level occurrence rate.

Figure 6 with 2 supplements see all

Download asset Open asset

First, we asked how many neural subsets were strongly correlated with each behavioral event and whether this count depended on behavioral event size. We found that many behavioral events had a significant number of such strongly correlated subsets and that the larger behavioral events tended to have a larger number of strongly correlated subsets (Figure 6B). Next, we asked how many behavioral events were strongly correlated with each neural subset and whether this count depended on the power-law range of the neural subset. We found that many subsets were strongly correlated with a significant number of behavioral events and that the subsets with greater power-law range tended to be correlated with more behavioral events (Figure 6C). The example results in Figure 6 are shown for all mice and all types of behaviors in Figure 6—figure supplements 1–2. Finally, we considered the time during which the mouse was active and asked what fraction of that time had a significant number of strongly correlated neural subsets. This fraction of time was often as high as 0.5 (Figure 6D). Thus, we conclude that the detailed, fast fluctuations of behavior are significantly and strongly correlated with scale-free neural activity.

Anticorrelated neural subsets

How is it possible for the total population to not exhibit scale-free fluctuations, while certain subsets are scale-free? A clue to this mystery is apparent upon close inspection of the results in Figure 4. Some neural subsets are strongly correlated with behavior, while other subsets are strongly anticorrelated, similar to previous reports from the motor cortex (Zagha et al., 2015) and prefrontal cortex (Garcia-Junco-Clemente et al., 2017). Since the scale-free subsets are often strongly correlated with behavior, it stands to reason that some neural subsets are strongly anti-correlated with the scale-free subsets. We show in Figure 7 that this fact explains why the total population does not exhibit scale-free dynamics. When two anticorrelated subsets are averaged together, they cancel out, resulting in relatively small fluctuations at the level of the total population (Figure 7A). The broad range of correlations and anticorrelations in each recording are more apparent using ‘correlation spectra’ (Figure 7B). For each previously defined seed neuron, we ranked all the other neurons according to their correlation with the seed neuron. These correlation spectra often reveal large numbers of strongly anticorrelated neurons. To directly show how cancelation of anticorrelated subsets abolishes scale-free dynamics, we re-computed the power-law range, but this time defined each subset to include the 25 most correlated and the 25 most anti-correlated neurons, relative to each seed neuron. The power-law range for these canceling subsets was greatly reduced compared to the original correlated subsets for all recordings (Figure 7C–E).

Figure 7

Download asset Open asset

Stochastic winner-take-all competition and criticality

What mechanisms might be responsible for scale-free dynamics among a certain subset of neurons that are obscured due to cancelation with anticorrelated subsets of neurons? Theoretical models for studying scale-free dynamics in neural networks have often employed randomly connected (Erdos-Renyi), networks of excitatory and inhibitory neurons, tuned to operate at criticality (Gautam et al., 2015; Li and Shew, 2020; Yang et al., 2012). However, such models do not manifest qualitatively different slow fluctuations in different subsets of neurons. More specifically, large groups of anticorrelated neurons do not occur in previous models. How can traditional models be changed to account for our observations? Here, we propose that a non-random network structure is needed. Indeed, experiments have shown that connectivity among cortical neurons is far from random (Cossell et al., 2015; Jiang et al., 2015; Song et al., 2005); this non-random structure is particularly pronounced among different types of inhibitory neurons (Jiang et al., 2015; Pfeffer et al., 2013; Pi et al., 2013; Wall et al., 2016). To our knowledge, this cell-type-specific connectivity has been ignored in previous models of scale-free neural activity.

Which aspects of this network structure might explain our findings? We initially sought clues from well-known circuit motifs that generate the anticorrelated activity underlying many types of animal locomotion (Kiehn, 2016). A common motif in these circuits, sometimes termed ‘crossing inhibition’ or ‘winner-take-all,’ entails two excitatory circuit nodes, say e+ and e−, whose activity is anticorrelated because they interact via at least one inhibitory node; e+ excites the inhibitory node, which suppresses e-, and vice versa, e- excites the inhibitory node which suppresses e+ (Figure 8A–D). Such circuit motifs certainly are common in the cortex, but they are mixed and interconnected with numerous other motifs. Even Erdos-Renyi networks contain many motifs of this type. Is it plausible that ‘winner-take-all’ motifs are responsible for our findings?

Figure 8 with 3 supplements see all

Download asset Open asset

As shown in Figure 8A, we tested this possibility using a computational model of 1000 binary neurons divided into two excitatory groups (e+ and e− with 400 neurons each) and two groups of inhibitory neurons (i+ and i− with 100 neurons each). We considered two inhibitory groups, instead of just one, to account for previous reports of anticorrelations between VIP and SOM inhibitory neurons in addition to anticorrelations between groups of excitatory neurons (Garcia-Junco-Clemente et al., 2017). The excitatory groups were densely connected within each group (50% connectivity) and sparsely connected across groups (5%), consistent with experimental observations (Cossell et al., 2015; Song et al., 2005). We also included dense crossing inhibition according to known connectivity among some inhibitory cell types (Jiang et al., 2015; Pfeffer et al., 2013; Pi et al., 2013; Wall et al., 2016). Excitatory and inhibitory synaptic weights were of the same absolute magnitude, but opposite in sign, and the connectivity matrix was normalized such that its largest eigenvalue was one (Figure 8B), following previous studies of scale-free dynamics and criticality (Gautam et al., 2015; Larremore et al., 2011; Li and Shew, 2020). Each neuron fired probabilistically, in proportion to the sum of its inputs (Methods).

We found that the e+ and e− neural subsets in this model produced large, slow, scale-free fluctuations (Figure 8C) like those observed in the scale-free subsets of neurons from our experimental observations (Figure 7A). The dynamics of e+ were strongly anticorrelated with those of e−, and i+ was anti-correlated with i− (Figure 8C), resulting in cancelation and a lack of scale-free dynamics when the entire population was considered together (Figure 8D). All neurons in the model were driven by unstructured random input; there was no statistical difference between the input to different subsets. Thus, the anticorrelated switching behavior in our model emerges stochastically. These results were robust to a wide range of different input rates (Figure 8E), provided that the largest eigenvalue of the connectivity matrix was near 1. If the largest eigenvalue was increased beyond one, the dynamics stopped switching between e+ and e−, instead getting locked into either e+ active and e− inactive or vice versa (this occurred to the right of the red dashed lines in Figure 8E and F). Thus, the stochastic switching between e+ and e− activity is a type of critical dynamics occurring at a boundary (red dashed line) in the model parameter space. Such critical winner-take-all dynamics never actually ‘choose a winner,’ instead switching randomly between the two ‘equally deserving winners.’ Scale-free anticorrelated fluctuations were also abolished when the density of connections between e+ and e− was increased beyond about 10% (Figure 8F). Above this density of connectivity between e+ and e−, the network effectively behaves like a single network, like previous models of criticality and scale-free dynamics.

In addition to testing dependence on the largest eigenvalue, the input rate, and the density of e+-to-e− connectivity, we did a systematic exploration of different network structures. We sought to understand if the network structure considered in Figure 8A–F is the only structure that reproduces our experimental observations, or if some other network structures could also work. Figure 8G summarizes the results of testing several other network structures. We found that crossing inhibition mediated by a single inhibitory population could suffice. The disinhibitory motif highlighted by other recent studies (Garcia-Junco-Clemente et al., 2017) is, thus, not necessary to reproduce our observations. Importantly, we showed that our observations can also occur in a more realistic network including a third population of inhibitory neurons with connectivity like that of PV neurons, according to known connectivity (Jiang et al., 2015; Pfeffer et al., 2013). We employed a brute force search considering 873,000 different combinations of dense, sparse, or zero connectivity for the 16 connections among the four network nodes in Figure 8A (more detail in Figure 8—figure supplement 1). We found only 31 configurations that matched the four criteria in Figure 8G. All matching configurations included one of two dense crossing inhibition motifs as well as segregated excitatory groups e+ and e− (Figure 8H and Figure 8—figure supplement 1). In fact, these two motifs were sufficient for predicting which configurations match our experimental results with 80% accuracy (Methods).

Finally, we considered the exponents and scaling relationships for the event size and duration distributions from the (example shown in Figure 9A, B and C). We considered the parts of model parameter space with non-zero power-law range for e+ (Figure 9D inset). We compared to experimental results from all neural subsets and all recordings (Figure 9E). For smaller power-law ranges, τ and α were more scattered and inconsistent. For the model, we found τ=1.3 ± 0.12, α=1.4 ± 0.15, and β=1.4 ± 0.04, for power-law range above three decades. For both the experiment and the model, the best-fit β was typically not well predicted by crackling noise theory; a confirmation of crackling noise theory would have the points falling on the unity line in the right panels of Figure 9D and E.

Figure 9

Download asset Open asset

While our model offers a simple explanation of anticorrelated scale-free dynamics, its simplicity comes with limitations. Perhaps the most obvious limitation of our model is that it does not include neurons with weak correlations to both e+ and e− (those neurons in the middle of the correlation spectrum shown in Figure 7B). In Figure 8—figure supplement 2, we show that our model can be modified in a simple way to include asynchronous neurons. Another limitation is that we assumed that all non-zero synaptic connections were equal in weight. We loosen this assumption allowing for variable weights in Figure 8—figure supplement 2, without changing the basic features of anticorrelated scale-free fluctuations. Future work might improve our model further by accounting for neurons with intermediate correlations, i.e., less sharp boundaries of the e+ and e− groups.

We have shown that ongoing, untrained locomotion, whisking, and pupil diameter changes in mice often exhibit scale-free dynamics. This scale-free behavior is directly related to concurrent scale-free cortical neural activity among certain subsets of neurons, with a strong one-to-one correspondence between many behavioral and neural events. Moreover, the subsets of neurons with the greatest range of scale-free-ness (i.e. the greatest power-law range) were the most correlated with behavior.

Why might a large power-law range among neurons be important for behavior? The behavioral variables we analyzed are likely to involve both local interactions among neighboring neurons and long-range coordination across many brain regions. For example, volitional movements like running and whisking presumably require coordinated interactions within and among motor, sensory, and other brain regions associated with decisions to move. One way to manifest such multiscale coordination from local to brain-wide circuits is to have neural events with sizes that occur with diverse scales, i.e., neural events with large power-law range.

Previous studies suggest additional functional implications of this close relationship between scale-free behavior and brain activity. First, scale-free neural activity has been associated with multiple functional benefits for sensory information processing and memory (Clawson et al., 2017; Gautam et al., 2015; Shew and Plenz, 2013; Shriki and Yellin, 2016; Suárez et al., 2021), while scale-free behavior has been associated with benefits for foraging, search, and decision making (Abe, 2020; Barabási, 2005; Garg and Kello, 2021; Sims et al., 2008; Sorribes et al., 2011; Viswanathan et al., 1999; Wosniack et al., 2017). Our results suggest that these two lists of functional benefits, previously considered separately, may in fact be a single unified list that occurs together. This long list of benefits may explain why several types of neural plasticity have been shown to maintain scale-free neural activity (Hoseini et al., 2017; Ma et al., 2019; Shew et al., 2015).

What mechanisms are responsible for scale-free neural activity? The prevailing view is the criticality hypothesis (see for example Beggs and Plenz, 2003; Muñoz, 2018; Shew and Plenz, 2013; Wilting and Priesemann, 2019). In this view, scale-free dynamics occur because the system operates near criticality, i.e., near the tipping point between two distinct dynamical regimes - one synchronous, the other asynchronous. However, previously studied models of criticality fail to explain several crucial components of our observations. First, we observed that scale-free activity is limited to subsets of neurons, which were strongly anticorrelated with other subsets of neurons. This apparent stochastic winner-take-all competition between subsets results in the cancelation of fluctuations at the level of the total population, thus obscuring the scale-free dynamics of the subsets. Previous models also do not agree with the power-law scaling relationships we found – especially our measurements of β in the relation S(T)~T^β. Our results suggest that a new type of criticality may be required to explain these observations. One possibility suggested by our model is that the scale-free dynamics we observe occur at the boundary between winner-less switching and single-winner locked-in dynamics (the red dashed line in Figure 8E and F). Additional theoretical efforts are necessary to more fully explore how the traditional criticality hypothesis relates to the competitive criticality suggested by our model.

At first glance, our results appear to not only contradict existing theory of criticality in neural systems, but also prior experimental observations based on different measurement methods. Most prominently, many previous studies have observed scale-free neural activity based on fast electrophysiological measurements using the total measured population, without accounting for anticorrelated subsets of neurons. Can fast ephys dynamics be reconciled with the slow calcium imaging signals we analyzed here? A detailed reconciliation of these different timescales of measurement will require more extensive investigation, but we provide a partial answer here (Figure 3—figure supplements 1–2). We analyzed an electrophysiological recording of spike times from 294 single units, also reported in the Stringer et al., 2019 paper. We found that when the activity is analyzed at a fast time scale (5 ms resolution), scale-free activity is clear using the full population and anticorrelations were very weak. The exponent of the power law was near –2, consistent with some previous reports based on time-resolved spike recordings in awake animals (Fontenele et al., 2019; Ma et al., 2019). In contrast, when the activity was analyzed at a slow time scale (330 ms resolution) like the calcium imaging data, strong anticorrelations emerged and scale-free dynamics were much more apparent (i.e. power law range was greater) when analyzed in subsets of neurons. This result (Figure 3—figure supplement 2) suggests that our findings are not inconsistent with prior studies of scale-free activity based on time-resolved electrophysiological data and calls for a more thorough study of how correlation structure and scale-free dynamics differ and relate across spatiotemporal scales. At the coarse time resolution of the calcium imaging data, we note that the neural subsets with large power-law range tended to be composed of neurons with high activity variance and high pairwise covariance (Figure 8—figure supplement 3). We note that a similar point about the emergence of anticorrelations at slow timescales was made in the original Stringer et al., 2019 paper.

What causes the scale-free fluctuations in behavior we observed from experiments? From the point of view of our model and the criticality hypothesis more generally, scale-free-ness originates in the dynamics of neural circuits. Indeed, the model generates scale-free fluctuations without any interactions with an ‘outside world.’ These internally generated scale-free neural dynamics then, might manifest as scale-free behavior. However, recalling that the experimental measurements reported here were from primary visual cortex, one might ask whether the scale-free fluctuations we observed in V1 are driven by scale-free fluctuations in visual input. In our view, this is unlikely given several facts. First, the gaze motion often had the least convincing scale-free-ness in our results (i.e. smallest power-law range, Figure 2C) and gaze motion would be the most obvious reason for changes in visual input (there were no dynamic visual stimuli on a viewed screen for these recordings). Second, the Stringer et al., paper (2019), from which our data originated, showed that spatially distributed electrophysiological recordings from multiple brain regions (not just sensory) were correlated to behavior in a similar way. This suggests that our results here may be a brain-wide phenomenon, not just a V1 phenomenon.

Setting aside speculations, our results firmly establish that complex fluctuations and events that make up scale-free neural activity are not ‘background noise’ nor ‘internal’ cognitive processes as is often supposed. Our findings call for a revision of these traditional views; scale-free neural activity directly causes (or perhaps is caused by) scale-free behavior and body motion.

The data analyzed here were first published in Stringer et al., 2019 and are publicly available on figshare at https://doi.org/10.25378/janelia.6163622.v6. Analysis code is publicly available on figshare at https://doi.org/10.1101/2021.05.12.443799.

1. 1. Shrew W

   (2023) figshare

   Code for computational model, power law fitting, power law range [Shew Lab].

   https://doi.org/10.6084/m9.figshare.21954389.v1

1. 1. Stringer C
   2. Pachitariu M
   3. Reddy C
   4. Carandini M
   5. Harris KD

   (2018) Figshare

   Recordings of ten thousand neurons in visual cortex during spontaneous behaviors.

   https://doi.org/10.25378/janelia.6163622.v6

1. 1. Abe MS

   (2020) Functional advantages of Lévy walks emerging near a critical point

   PNAS 117:24336–24344.

   https://doi.org/10.1073/pnas.2001548117

   • PubMed
   • Google Scholar
2. 1. Anteneodo C
   2. Chialvo DR

   (2009) Unraveling the fluctuations of animal motor activity

   Chaos 19:033123.

   https://doi.org/10.1063/1.3211189

   • PubMed
   • Google Scholar
3. 1. Barabási AL

   (2005) The origin of bursts and heavy tails in human dynamics

   Nature 435:207–211.

   https://doi.org/10.1038/nature03459

   • PubMed
   • Google Scholar
4. 1. Beggs JM
   2. Plenz D

   (2003) Neuronal avalanches in neocortical circuits

   The Journal of Neuroscience 23:11167–11177.

   https://doi.org/10.1523/JNEUROSCI.23-35-11167.2003

   • PubMed
   • Google Scholar
5. 1. Bellay T
   2. Klaus A
   3. Seshadri S
   4. Plenz D

   (2015) Irregular spiking of pyramidal neurons organizes as scale-invariant neuronal avalanches in the awake state

   eLife 4:e07224.

   https://doi.org/10.7554/eLife.07224

   • PubMed
   • Google Scholar
6. 1. Bellay T
   2. Shew WL
   3. Yu S
   4. Falco-Walter JJ
   5. Plenz D

   (2020) Selective participation of single cortical neurons in neuronal avalanches

   Frontiers in Neural Circuits 14:620052.

   https://doi.org/10.3389/fncir.2020.620052

   • PubMed
   • Google Scholar
7. 1. Clancy KB
   2. Orsolic I
   3. Mrsic-Flogel TD

   (2019) Locomotion-dependent remapping of distributed cortical networks

   Nature Neuroscience 22:778–786.

   https://doi.org/10.1038/s41593-019-0357-8

   • PubMed
   • Google Scholar
8. 1. Clauset A
   2. Shalizi CR
   3. Newman MEJ

   (2009) Power-Law distributions in empirical data

   SIAM Review 51:661–703.

   https://doi.org/10.1137/070710111

   • Google Scholar
9. 1. Clawson WP
   2. Wright NC
   3. Wessel R
   4. Shew WL

   (2017) Adaptation towards scale-free dynamics improves cortical stimulus discrimination at the cost of reduced detection

   PLOS Computational Biology 13:e1005574.

   https://doi.org/10.1371/journal.pcbi.1005574

   • PubMed
   • Google Scholar
10. 1. Cossell L
    2. Iacaruso MF
    3. Muir DR
    4. Houlton R
    5. Sader EN
    6. Ko H
    7. Hofer SB
    8. Mrsic-Flogel TD

    (2015) Functional organization of excitatory synaptic strength in primary visual cortex

    Nature 518:399–403.

    https://doi.org/10.1038/nature14182

    • PubMed
    • Google Scholar
11. 1. Dalla Porta L
    2. Copelli M

    (2019) Modeling neuronal avalanches and long-range temporal correlations at the emergence of collective oscillations: continuously varying exponents mimic M/EEG results

    PLOS Computational Biology 15:e1006924.

    https://doi.org/10.1371/journal.pcbi.1006924

    • PubMed
    • Google Scholar
12. 1. di Santo S
    2. Villegas P
    3. Burioni R
    4. Muñoz MA

    (2018) Landau-ginzburg theory of cortex dynamics: scale-free avalanches emerge at the edge of synchronization

    PNAS 115:E1356–E1365.

    https://doi.org/10.1073/pnas.1712989115

    • PubMed
    • Google Scholar
13. 1. Fontenele AJ
    2. de Vasconcelos NAP
    3. Feliciano T
    4. Aguiar LAA
    5. Soares-Cunha C
    6. Coimbra B
    7. Dalla Porta L
    8. Ribeiro S
    9. Rodrigues AJ
    10. Sousa N
    11. Carelli PV
    12. Copelli M

    (2019) Criticality between cortical states

    Physical Review Letters 122:208101.

    https://doi.org/10.1103/PhysRevLett.122.208101

    • PubMed
    • Google Scholar
14. 1. Fosque LJ
    2. Williams-García RV
    3. Beggs JM
    4. Ortiz G

    (2021) Evidence for quasicritical brain dynamics

    Physical Review Letters 126:98101.

    https://doi.org/10.1103/PhysRevLett.126.098101

    • Google Scholar
15. 1. Friedman N
    2. Ito S
    3. Brinkman BAW
    4. Shimono M
    5. DeVille REL
    6. Dahmen KA
    7. Beggs JM
    8. Butler TC

    (2012) Universal critical dynamics in high resolution neuronal avalanche data

    Physical Review Letters 108:1–5.

    https://doi.org/10.1103/PhysRevLett.108.208102

    • PubMed
    • Google Scholar
16. 1. Friedrich J
    2. Zhou P
    3. Paninski L

    (2017) Fast online deconvolution of calcium imaging data

    PLOS Computational Biology 13:e1005423.

    https://doi.org/10.1371/journal.pcbi.1005423

    • PubMed
    • Google Scholar
17. 1. Garcia-Junco-Clemente P
    2. Ikrar T
    3. Tring E
    4. Xu X
    5. Ringach DL
    6. Trachtenberg JT

    (2017) An inhibitory pull-push circuit in frontal cortex

    Nature Neuroscience 20:389–392.

    https://doi.org/10.1038/nn.4483

    • PubMed
    • Google Scholar
18. 1. Garg K
    2. Kello CT

    (2021) Efficient Lévy walks in virtual human foraging

    Scientific Reports 11:5242.

    https://doi.org/10.1038/s41598-021-84542-w

    • PubMed
    • Google Scholar
19. 1. Gautam SH
    2. Hoang TT
    3. McClanahan K
    4. Grady SK
    5. Shew WL

    (2015) Maximizing sensory dynamic range by tuning the cortical state to criticality

    PLOS Computational Biology 11:e1004576.

    https://doi.org/10.1371/journal.pcbi.1004576

    • PubMed
    • Google Scholar
20. 1. Girardi-Schappo M
    2. Brochini L
    3. Costa AA
    4. Carvalho TTA
    5. Kinouchi O

    (2020) Synaptic balance due to homeostatically self-organized quasicritical dynamics

    Physical Review Research 2:12042.

    https://doi.org/10.1103/PhysRevResearch.2.012042

    • Google Scholar
21. 1. Gireesh ED
    2. Plenz D

    (2008) Neuronal avalanches organize as nested theta- and beta/gamma-oscillations during development of cortical layer 2/3

    PNAS 105:7576–7581.

    https://doi.org/10.1073/pnas.0800537105

    • PubMed
    • Google Scholar
22. 1. Hardstone R
    2. Poil S-S
    3. Schiavone G
    4. Jansen R
    5. Nikulin VV
    6. Mansvelder HD
    7. Linkenkaer-Hansen K

    (2012) Detrended fluctuation analysis: a scale-free view on neuronal oscillations

    Frontiers in Physiology 3:450.

    https://doi.org/10.3389/fphys.2012.00450

    • PubMed
    • Google Scholar
23. 1. Hoseini MS
    2. Pobst J
    3. Wright NC
    4. Clawson W
    5. Shew W
    6. Wessel R

    (2017) Induced cortical oscillations in turtle cortex are coherent at the mesoscale of population activity, but not at the microscale of the membrane potential of neurons

    Journal of Neurophysiology 118:2579–2591.

    https://doi.org/10.1152/jn.00375.2017

    • PubMed
    • Google Scholar
24. 1. Huang L
    2. Ledochowitsch P
    3. Knoblich U
    4. Lecoq J
    5. Murphy GJ
    6. Reid RC
    7. de Vries SE
    8. Koch C
    9. Zeng H
    10. Buice MA
    11. Waters J
    12. Li L

    (2021) Relationship between simultaneously recorded spiking activity and fluorescence signal in gcamp6 transgenic mice

    eLife 10:e51675.

    https://doi.org/10.7554/eLife.51675

    • PubMed
    • Google Scholar
25. 1. Jiang X
    2. Shen S
    3. Cadwell CR
    4. Berens P
    5. Sinz F
    6. Ecker AS
    7. Patel S
    8. Tolias AS

    (2015) Principles of connectivity among morphologically defined cell types in adult neocortex

    Science 350:aac9462.

    https://doi.org/10.1126/science.aac9462

    • PubMed
    • Google Scholar
26. 1. Kello CT
    2. Brown GDA
    3. Ferrer-I-Cancho R
    4. Holden JG
    5. Linkenkaer-Hansen K
    6. Rhodes T
    7. Van Orden GC

    (2010) Scaling laws in cognitive sciences

    Trends in Cognitive Sciences 14:223–232.

    https://doi.org/10.1016/j.tics.2010.02.005

    • PubMed
    • Google Scholar
27. 1. Kiehn O

    (2016) Decoding the organization of spinal circuits that control locomotion

    Nature Reviews. Neuroscience 17:224–238.

    https://doi.org/10.1038/nrn.2016.9

    • PubMed
    • Google Scholar
28. 1. Langlois D
    2. Cousineau D
    3. Thivierge JP

    (2014) Maximum likelihood estimators for truncated and censored power-law distributions show how neuronal avalanches may be misevaluated

    Physical Review. E, Statistical, Nonlinear, and Soft Matter Physics 89:012709.

    https://doi.org/10.1103/PhysRevE.89.012709

    • PubMed
    • Google Scholar
29. 1. Larremore DB
    2. Shew WL
    3. Restrepo JG

    (2011) Predicting criticality and dynamic range in complex networks: effects of topology

    Physical Review Letters 106:1–4.

    https://doi.org/10.1103/PhysRevLett.106.058101

    • PubMed
    • Google Scholar
30. 1. Larremore DB
    2. Shew WL
    3. Ott E
    4. Sorrentino F
    5. Restrepo JG

    (2014) Inhibition causes ceaseless dynamics in networks of excitable nodes

    Physical Review Letters 112:138103.

    https://doi.org/10.1103/PhysRevLett.112.138103

    • PubMed
    • Google Scholar
31. 1. Li J
    2. Shew WL

    (2020) Tuning network dynamics from criticality to an asynchronous state

    PLOS Computational Biology 16:e1008268.

    https://doi.org/10.1371/journal.pcbi.1008268

    • PubMed
    • Google Scholar
32. 1. Ma Z
    2. Turrigiano GG
    3. Wessel R
    4. Hengen KB

    (2019) Cortical circuit dynamics are homeostatically tuned to criticality in vivo

    Neuron 104:655–664.

    https://doi.org/10.1016/j.neuron.2019.08.031

    • PubMed
    • Google Scholar
33. 1. Muñoz MA

    (2018) Colloquium: criticality and dynamical scaling in living systems

    Reviews of Modern Physics 90:31001.

    https://doi.org/10.1093/ejcts/ezx068

    • PubMed
    • Google Scholar
34. 1. Musall S
    2. Kaufman MT
    3. Juavinett AL
    4. Gluf S
    5. Churchland AK

    (2019) Single-trial neural dynamics are dominated by richly varied movements

    Nature Neuroscience 22:1677–1686.

    https://doi.org/10.1038/s41593-019-0502-4

    • PubMed
    • Google Scholar
35. 1. Palva JM
    2. Zhigalov A
    3. Hirvonen J
    4. Korhonen O
    5. Linkenkaer-Hansen K
    6. Palva S

    (2013) Neuronal long-range temporal correlations and avalanche dynamics are correlated with behavioral scaling laws

    PNAS 110:3585–3590.

    https://doi.org/10.1073/pnas.1216855110

    • PubMed
    • Google Scholar
36. 1. Pfeffer CK
    2. Xue M
    3. He M
    4. Huang ZJ
    5. Scanziani M

    (2013) Inhibition of inhibition in visual cortex: the logic of connections between molecularly distinct interneurons

    Nature Neuroscience 16:1068–1076.

    https://doi.org/10.1038/nn.3446

    • PubMed
    • Google Scholar
37. 1. Pi HJ
    2. Hangya B
    3. Kvitsiani D
    4. Sanders JI
    5. Huang ZJ
    6. Kepecs A

    (2013) Cortical interneurons that specialize in disinhibitory control

    Nature 503:521–524.

    https://doi.org/10.1038/nature12676

    • PubMed
    • Google Scholar
38. 1. Poil SS
    2. Hardstone R
    3. Mansvelder HD
    4. Linkenkaer-Hansen K

    (2012) Critical-state dynamics of avalanches and oscillations jointly emerge from balanced excitation/inhibition in neuronal networks

    The Journal of Neuroscience 32:9817–9823.

    https://doi.org/10.1523/JNEUROSCI.5990-11.2012

    • PubMed
    • Google Scholar
39. 1. Priesemann V
    2. Wibral M
    3. Valderrama M
    4. Pröpper R
    5. Le Van Quyen M
    6. Geisel T
    7. Triesch J
    8. Nikolić D
    9. Munk MHJ

    (2014) Spike avalanches in vivo suggest a driven, slightly subcritical brain state

    Frontiers in Systems Neuroscience 8:108.

    https://doi.org/10.3389/fnsys.2014.00108

    • PubMed
    • Google Scholar
40. 1. Proekt A
    2. Banavar JR
    3. Maritan A
    4. Pfaff DW

    (2012) Scale invariance in the dynamics of spontaneous behavior

    PNAS 109:10564–10569.

    https://doi.org/10.1073/pnas.1206894109

    • PubMed
    • Google Scholar
41. 1. Reimer J
    2. Froudarakis E
    3. Cadwell CR
    4. Yatsenko D
    5. Denfield GH
    6. Tolias AS

    (2014) Pupil fluctuations track fast switching of cortical states during quiet wakefulness

    Neuron 84:355–362.

    https://doi.org/10.1016/j.neuron.2014.09.033

    • PubMed
    • Google Scholar
42. 1. Reimer J
    2. McGinley MJ
    3. Liu Y
    4. Rodenkirch C
    5. Wang Q
    6. McCormick DA
    7. Tolias AS

    (2016) Pupil fluctuations track rapid changes in adrenergic and cholinergic activity in cortex

    Nature Communications 7:13289.

    https://doi.org/10.1038/ncomms13289

    • PubMed
    • Google Scholar
43. 1. Salkoff DB
    2. Zagha E
    3. McCarthy E
    4. McCormick DA

    (2020) Movement and performance explain widespread cortical activity in a visual detection task

    Cerebral Cortex 30:421–437.

    https://doi.org/10.1093/cercor/bhz206

    • PubMed
    • Google Scholar
44. 1. Scott JW
    2. Sherrill L
    3. Jiang J
    4. Zhao K

    (2014) Tuning to odor solubility and sorption pattern in olfactory epithelial responses

    The Journal of Neuroscience 34:2025–2036.

    https://doi.org/10.1523/JNEUROSCI.3736-13.2014

    • PubMed
    • Google Scholar
45. 1. Sethna JP
    2. Dahmen KA
    3. Myers CR

    (2001) Crackling noise

    Nature 410:242–250.

    https://doi.org/10.1038/35065675

    • PubMed
    • Google Scholar
46. 1. Shew WL
    2. Yang H
    3. Petermann T
    4. Roy R
    5. Plenz D

    (2009) Neuronal avalanches imply maximum dynamic range in cortical networks at criticality

    The Journal of Neuroscience 29:15595–15600.

    https://doi.org/10.1523/JNEUROSCI.3864-09.2009

    • PubMed
    • Google Scholar
47. 1. Shew WL
    2. Plenz D

    (2013) The functional benefits of criticality in the cortex

    The Neuroscientist 19:88–100.

    https://doi.org/10.1177/1073858412445487

    • PubMed
    • Google Scholar
48. 1. Shew WL
    2. Clawson WP
    3. Pobst J
    4. Karimipanah Y
    5. Wright NC
    6. Wessel R

    (2015) Adaptation to sensory input tunes visual cortex to criticality

    Nature Physics 11:659–663.

    https://doi.org/10.1038/nphys3370

    • Google Scholar
49. 1. Shriki O
    2. Alstott J
    3. Carver F
    4. Holroyd T
    5. Henson RNA
    6. Smith ML
    7. Coppola R
    8. Bullmore E
    9. Plenz D

    (2013) Neuronal avalanches in the resting MEG of the human brain

    The Journal of Neuroscience 33:7079–7090.

    https://doi.org/10.1523/JNEUROSCI.4286-12.2013

    • PubMed
    • Google Scholar
50. 1. Shriki O
    2. Yellin D

    (2016) Optimal information representation and criticality in an adaptive sensory recurrent neuronal network

    PLOS Computational Biology 12:e1004698.

    https://doi.org/10.1371/journal.pcbi.1004698

    • PubMed
    • Google Scholar
51. 1. Sims DW
    2. Southall EJ
    3. Humphries NE
    4. Hays GC
    5. Bradshaw CJA
    6. Pitchford JW
    7. James A
    8. Ahmed MZ
    9. Brierley AS
    10. Hindell MA
    11. Morritt D
    12. Musyl MK
    13. Righton D
    14. Shepard ELC
    15. Wearmouth VJ
    16. Wilson RP
    17. Witt MJ
    18. Metcalfe JD

    (2008) Scaling laws of marine predator search behaviour

    Nature 451:1098–1102.

    https://doi.org/10.1038/nature06518

    • PubMed
    • Google Scholar
52. 1. Smit DJA
    2. Linkenkaer-Hansen K
    3. de Geus EJC

    (2013) Long-range temporal correlations in resting-state α oscillations predict human timing-error dynamics

    The Journal of Neuroscience 33:11212–11220.

    https://doi.org/10.1523/JNEUROSCI.2816-12.2013

    • PubMed
    • Google Scholar
53. 1. Song S
    2. Sjöström PJ
    3. Reigl M
    4. Nelson S
    5. Chklovskii DB

    (2005) Highly nonrandom features of synaptic connectivity in local cortical circuits

    PLOS Biology 3:e68.

    https://doi.org/10.1371/journal.pbio.0030068

    • PubMed
    • Google Scholar
54. 1. Sorribes A
    2. Armendariz BG
    3. Lopez-Pigozzi D
    4. Murga C
    5. de Polavieja GG

    (2011) The origin of behavioral bursts in decision-making circuitry

    PLOS Computational Biology 7:e1002075.

    https://doi.org/10.1371/journal.pcbi.1002075

    • PubMed
    • Google Scholar
55. 1. Stringer C
    2. Pachitariu M
    3. Steinmetz N
    4. Reddy CB
    5. Carandini M
    6. Harris KD

    (2019) Spontaneous behaviors drive multidimensional, brainwide activity

    Science 364:255.

    https://doi.org/10.1126/science.aav7893

    • PubMed
    • Google Scholar
56. 1. Suárez LE
    2. Richards BA
    3. Lajoie G
    4. Misic B

    (2021) Learning function from structure in neuromorphic networks

    Nature Machine Intelligence 3:771–786.

    https://doi.org/10.1038/s42256-021-00376-1

    • Google Scholar
57. 1. Tagliazucchi E
    2. Balenzuela P
    3. Fraiman D
    4. Chialvo DR

    (2012) Criticality in large-scale brain FMRI dynamics unveiled by a novel point process analysis

    Frontiers in Physiology 3:15.

    https://doi.org/10.3389/fphys.2012.00015

    • PubMed
    • Google Scholar
58. 1. Tetzlaff C
    2. Okujeni S
    3. Egert U
    4. Wörgötter F
    5. Butz M

    (2010) Self-organized criticality in developing neuronal networks

    PLOS Computational Biology 6:e1001013.

    https://doi.org/10.1371/journal.pcbi.1001013

    • PubMed
    • Google Scholar
59. 1. Vinck M
    2. Batista-Brito R
    3. Knoblich U
    4. Cardin JA

    (2015) Arousal and locomotion make distinct contributions to cortical activity patterns and visual encoding

    Neuron 86:740–754.

    https://doi.org/10.1016/j.neuron.2015.03.028

    • PubMed
    • Google Scholar
60. 1. Viswanathan GM
    2. Buldyrev SV
    3. Havlin S
    4. da Luz MG
    5. Raposo EP
    6. Stanley HE

    (1999) Optimizing the success of random searches

    Nature 401:911–914.

    https://doi.org/10.1038/44831

    • PubMed
    • Google Scholar
61. 1. Wall NR
    2. De La Parra M
    3. Sorokin JM
    4. Taniguchi H
    5. Huang ZJ
    6. Callaway EM

    (2016) Brain-Wide maps of synaptic input to cortical interneurons

    The Journal of Neuroscience 36:4000–4009.

    https://doi.org/10.1523/JNEUROSCI.3967-15.2016

    • PubMed
    • Google Scholar
62. 1. Wilting J
    2. Dehning J
    3. Pinheiro Neto J
    4. Rudelt L
    5. Wibral M
    6. Zierenberg J
    7. Priesemann V

    (2018) Operating in a reverberating regime enables rapid tuning of network states to task requirements

    Frontiers in Systems Neuroscience 12:55.

    https://doi.org/10.3389/fnsys.2018.00055

    • PubMed
    • Google Scholar
63. 1. Wilting J
    2. Priesemann V

    (2019) 25 years of criticality in neuroscience-established results, open controversies, novel concepts

    Current Opinion in Neurobiology 58:105–111.

    https://doi.org/10.1016/j.conb.2019.08.002

    • PubMed
    • Google Scholar
64. 1. Wosniack ME
    2. Santos MC
    3. Raposo EP
    4. Viswanathan GM
    5. da Luz MGE

    (2017) The evolutionary origins of Lévy walk foraging

    PLOS Computational Biology 13:e1005774.

    https://doi.org/10.1371/journal.pcbi.1005774

    • PubMed
    • Google Scholar
65. 1. Yang H
    2. Shew WL
    3. Roy R
    4. Plenz D

    (2012) Maximal variability of phase synchrony in cortical networks with neuronal avalanches

    The Journal of Neuroscience 32:1061–1072.

    https://doi.org/10.1523/JNEUROSCI.2771-11.2012

    • PubMed
    • Google Scholar
66. 1. Yu S
    2. Ribeiro TL
    3. Meisel C
    4. Chou S
    5. Mitz A
    6. Saunders R
    7. Plenz D

    (2017) Maintained avalanche dynamics during task-induced changes of neuronal activity in nonhuman primates

    eLife 6:e27119.

    https://doi.org/10.7554/eLife.27119

    • PubMed
    • Google Scholar
67. 1. Zagha E
    2. Ge X
    3. McCormick DA

    (2015) Competing neural ensembles in motor cortex gate goal-directed motor output

    Neuron 88:565–577.

    https://doi.org/10.1016/j.neuron.2015.09.044

    • PubMed
    • Google Scholar

Author details

1. Sabrina A Jones

   Department of Physics, University of Arkansas at Fayetteville, Fayetteville, United States

   Contribution

   Conceptualization, Software, Formal analysis, Investigation, Methodology, Writing – review and editing

   Competing interests

   No competing interests declared

2. Jacob H Barfield

   Department of Physics, University of Arkansas at Fayetteville, Fayetteville, United States

   Contribution

   Software, Formal analysis, Investigation, Methodology, Writing – review and editing

   Competing interests

   No competing interests declared

3. V Kindler Norman

   Department of Physics, University of Arkansas at Fayetteville, Fayetteville, United States

   Contribution

   Software, Formal analysis, Investigation

   Competing interests

   No competing interests declared

4. Woodrow L Shew

   Department of Physics, University of Arkansas at Fayetteville, Fayetteville, United States

   Contribution

   Conceptualization, Software, Formal analysis, Supervision, Funding acquisition, Investigation, Methodology, Writing – original draft, Project administration, Writing – review and editing

   For correspondence

   shew@uark.edu

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0003-0679-1766

• 2,298

  views

• 373

  downloads

• 16

  citations

Views, downloads and citations are aggregated across all versions of this paper published by eLife.