Efficient “Communication through Coherence” Requires Oscillations Structured to Minimize Interference between Signals

“Such activity spans several orders of magnitude in frequency [31], and in several brain regions the phase of firing is actively modulated relative to a single coherent oscillation [26,28,29].”

“A possible function of such activity is to modulate functional connectivity among anatomically connected regions [4–6].”

“This study provides a quantitative assessment of the proposal that selective communication can be achieved by coherence between firing rate modulation in a sending region and gain modulation in a receiving region [5].”

“The resulting activity was consistent with in vivo data showing irregular spiking of single units [16–18] during sparsely synchronized oscillatory activity [19].”

“The ‘communication through coherence’ (CTC) hypothesis [5,7] proposes that selective communication is achieved through coherence between firing rate oscillation in the sending region and oscillatory gain modulation in the receiving region.”

“…different oscillations, from very strong modulations during hippocampal theta oscillations [47], and oscillations in the olfactory system of zebrafish [27] and locusts [48], to apparently weaker modulation in some studies of gamma oscillations in the hippocampus [49,50] and entorhinal cortex [49].”

“We next evaluated the proposal that changes in the inter-region coherence of oscillatory activity [5,8,10] (as distinct from changes in frequency or amplitude, or changes in a consistent phase relationship) could be used to switch on or off information propagation through a convergent pathway.”

“…show a wide range of modulation strengths across different oscillations, from very strong modulations during hippocampal theta oscillations [47], and oscillations in the olfactory system of zebrafish [27] and locusts [48], to apparently weaker modulation in some studies of gamma…”

“A possible function of such activity is to modulate functional connectivity among anatomically connected regions [4–6].”

“Changes in oscillatory activity dependent on visual spatial attention [3,16,40,41] have been proposed to underlie the selective processing of behaviorally relevant stimuli [5,7].”

“We note that gamma oscillations in V1 are particularly amenable to experimental phase manipulation as they are readily entrained by flickering visual stimuli [43], as expected given the response dynamics of gamma oscillating networks in vitro [44].”

“Various studies have reported switching between distinct oscillation frequencies in a local network [36–38], potentially reflecting participation in distinct large scale networks utilizing different frequencies for communication.”

“Task-dependent changes in the power and inter-region coherence of oscillatory network activity are observed in many brain regions and behavioral tasks [1–3].”

“Striking bursts of transient task dependent oscillatory activity are well documented in many brain regions including in motor cortex during movement preparation [32,33], the basal ganglia during cue utilization [34], and in visual cortex during working memory [35].”

“Despite these technical difficulties reported spike phase histograms show a wide range of modulation strengths across different oscillations, from very strong modulations during hippocampal theta oscillations [47], and oscillations in the olfactory system of zebrafish [27] and locusts [48], to apparently weaker modula…”

“Under many noise distributions these decoders perform close to optimally, as indicated by the minimal difference in performance when compared with more sophisticated non-linear methods [22,23].”

“Physiologically, gain modulation could be achieved by local interneuron circuitry modulating the distribution of membrane potentials, degree of shunting inhibition [20], or synaptic noise [21] experienced by the principal neurons of the receiving network.”

“Oscillatory activity can also modulate the gain of synaptic plasticity [54,55], and spike timing dependent plasticity with an oscillating post synaptic population will also produce periodic modulation of the gain for plasticity.”

“Striking bursts of transient task dependent oscillatory activity are well documented in many brain regions including in motor cortex during movement preparation [32,33], the basal ganglia during cue utilization [34], and in visual cortex during working memory [35].”

“We next evaluated the proposal that changes in the inter-region coherence of oscillatory activity [5,8,10] (as distinct from changes in frequency or amplitude, or changes in a consistent phase relationship) could be used to switch on or off information propagation through a convergent pathway.”

“Changes in oscillatory activity dependent on visual spatial attention [3,16,40,41] have been proposed to underlie the selective processing of behaviorally relevant stimuli [5,7].”

“Oscillatory activity can also modulate the gain of synaptic plasticity [54,55], and spike timing dependent plasticity with an oscillating post synaptic population will also produce periodic modulation of the gain for plasticity.”

“Task-dependent changes in the power and inter-region coherence of oscillatory network activity are observed in many brain regions and behavioral tasks [1–3].”

“These decoders are, moreover, biologically plausible as their performance corresponds to that of de-noising by networks implementing line attractor dynamics [24,25].”

“Striking bursts of transient task dependent oscillatory activity are well documented in many brain regions including in motor cortex during movement preparation [32,33], the basal ganglia during cue utilization [34], and in visual cortex during working memory [35].”

“…different oscillations, from very strong modulations during hippocampal theta oscillations [47], and oscillations in the olfactory system of zebrafish [27] and locusts [48], to apparently weaker modulation in some studies of gamma oscillations in the hippocampus [49,50] and entorhinal cortex [49].”

“Striking bursts of transient task dependent oscillatory activity are well documented in many brain regions including in motor cortex during movement preparation [32,33], the basal ganglia during cue utilization [34], and in visual cortex during working memory [35].”

“…different oscillations, from very strong modulations during hippocampal theta oscillations [47], and oscillations in the olfactory system of zebrafish [27] and locusts [48], to apparently weaker modulation in some studies of gamma oscillations in the hippocampus [49,50] and entorhinal cortex [49].”

“We previously developed a biophysical model [15], which exploits network resonance effects at the boundary between asynchronous and oscillating states [51] to selectively respond to inputs oscillating at a specific frequency.”

“Under many noise distributions these decoders perform close to optimally, as indicated by the minimal difference in performance when compared with more sophisticated non-linear methods [22,23].”

“Such activity spans several orders of magnitude in frequency [31], and in several brain regions the phase of firing is actively modulated relative to a single coherent oscillation [26,28,29].”

“One possibility discussed by the authors is that theta frequency resetting of gamma oscillation phase [42] across V1 and V4, combined with the frequency offset between the V1 networks, creates periods in which the attended V1 site consistently leads the unattended site.”

“Various studies have reported switching between distinct oscillation frequencies in a local network [36–38], potentially reflecting participation in distinct large scale networks utilizing different frequencies for communication.”

“Despite extensive experimental [8–10] and computational work [11–14], it remains unclear under what conditions the CTC mechanism could allow a network to distinguish among converging population-coded signals, and how its performance depends on the structure of their oscillatory modulations.”

“We previously developed a biophysical model [15], which exploits network resonance effects at the boundary between asynchronous and oscillating states [51] to selectively respond to inputs oscillating at a specific frequency.”

“The resulting activity was consistent with in vivo data showing irregular spiking of single units [16–18] during sparsely synchronized oscillatory activity [19].”

“We note that gamma oscillations in V1 are particularly amenable to experimental phase manipulation as they are readily entrained by flickering visual stimuli [43], as expected given the response dynamics of gamma oscillating networks in vitro [44].”

“Several studies working in this direction [12,14,52,53] have demonstrated some degree of input selectivity on the basis of modulation, particularly for phase separated inputs [52,53].”

“The widely used measures of spike-field and spike-spike coherence do not map directly onto modulation strength as they are confounded by firing rate [46], which substantially impedes attempts to evaluate modulation strength from much of the published literature.”

“These decoders are, moreover, biologically plausible as their performance corresponds to that of de-noising by networks implementing line attractor dynamics [24,25].”

“Despite extensive experimental [8–10] and computational work [11–14], it remains unclear under what conditions the CTC mechanism could allow a network to distinguish among converging population-coded signals, and how its performance depends on the structure of their oscillatory modulations.”

“Several studies working in this direction [12,14,52,53] have demonstrated some degree of input selectivity on the basis of modulation, particularly for phase separated inputs [52,53].”

“Several studies working in this direction [12,14,52,53] have demonstrated some degree of input selectivity on the basis of modulation, particularly for phase separated inputs [52,53].”

“A recent study found no effect of manipulating the relative phase of gamma frequency flicker between target and distracting stimuli on selective attentional processing [45], although without concurrent electrophysiological data it is unclear how effectively cortical activity was manipulated.”

“Mathematical implementations of CTC that assume a sinusoidal oscillation and a linear relation between phase and excitability show that presynaptic groups, which are incoherent to the postsynaptic group, might still have a substantial impact (Akam and Kullmann, 2012).”

“Computational studies suggest that strong coherence is required for selective routing (Akam and Kullmann, 2012).”

“…ranging from single-neuron activity as subthreshold membrane potential oscillations and action potentials, to local activity of assemblies of neurons, and even to activity patterns of whole cortical networks in context of different brain areas (e.g., Akam and Kullmann, 2012; Buzsáki et al., 2013).”

“Does speci fi c long-distance communication improve when local gamma oscillations are locked to global slow oscillations [99]?”

“…interregional communication (Singer, 1993; Fries, 2005), this hypothesis remains hotly debated (Shadlen and Movshon, 1999; Kopell et al., 2000; Fries, 2009; Ray and Maunsell, 2010; Fell and Axmacher, 2011; Bosman et al., 2012; Akam and Kullmann, 2012; Jia et al., 2013; Srinath and Ray, 2014).”

“Another approach to coordination of sender and target was taken by Akam & Kullmann (2012).”

“Computational modeling demonstrates that selective communication can indeed be achieved by coherent oscillatory gain modulation, but that the structure of oscillatory activity must satisfy certain constraints (Akam and Kullmann, 2012).”

“Selective gating of signals through network resonance has been suggested by previous theoretical studies [51,67].”

“(ii) Later work considered scenarios in which multiple neuronal groups have a gamma-rhythm, and examined how the integration of the synchronous input depends on the ongoing activity of the receiver (Fries, 2005, 2015; Börgers et al., 2008; Gielen et al., 2010; Akam and Kullmann, 2012).”

“Gamma synchrony has been reported both in subcortical structures (Akam & Kullmann, 2012; Steriade et al., 1993; Zhou et al., 2016) and in cortical areas (Fries, 2015; Gray & Singer, 1989; Gregoriou et al., 2009).”

“neurophysiological network over which the brain processes information during cognitive processes [1,2].”

“CTC has received support from modeling studies (Akam and Kullmann, 2010, 2012) and experimental results (Saalmann et al., 2012; Roberts et al., 2013).”

“…depends on the relative gamma phase (zero phase) between the spikes and the LFP. Similar to the situation with the BBS hypothesis, the CTC hypothesis was both supported (Womelsdorf et al. 2007; Gregoriou et al. 2009) and opposed (Jia et al. 2013a, Akam and Kullmann 2012).”

“Given such variability, it remains an open problem how attention can be generically mediated via selective interareal synchronization, and how stimuli with weak g-synchronization are communicated to higher areas (for further discussion see Hermes et al. 2015; Ray and Maunsell 2015; Brunet et al. 2014; Akam and Kullman…”

“Among other functions, oscillations are thought to provide a temporal structure for local and long-range neural communication in the brain (Akam and Kullmann, 2012; Buzsáki and Moser, 2013; Fries, 2005).”

“Recent computational studies further suggest that selective communication is achieved by coherence mechanisms between gain (signal to noise ratio, amplitude scaling) and firing rate modulations (Akam and Kullmann, 2012).”

“(5) Longer gamma cycles were associated with stronger spikefield phase-locking (synchrony), but lower firing-rates, and were not accompanied by stronger, transient spiking activation.”

“Frontiers in Human Neuroscience www.frontiersin.org September 2013 | Volume 7 | Article 614 | 2 as the interplay between intrinsic properties of the neuron and its inputs, are matters of ongoing research (Akam and Kullmann, 2012).”

“As a metric of functional coupling, correlations of amplitudes, as well as parameters measuring the extent of phase-locked activity have been introduced [Akam and Kullmann, 2012; Bastos et al., 2015; Fries, 2005, 2015].”

“Synchronized oscillations in neural circuits could provide, at least theoretically, a universal device for pattern recognition and information handling in the brain [1,2].”

“Understanding the 60 mechanisms mediating the coexistence of these rhythms, as well as state-dependent changes 61 in their properties, would yield important insight about how collective neural activity and 62 synchronization phenomena, shaped by both sensory and recurrent inputs, mediate neural 63 communication[3].”

“Models have also identified some limits and requirements for an synchrony-based dynamic communication structure (Akam and Kullmann, 2012; Rolls et al., 2012).”

“These observations exemplify the communicationthroughcoherence theory (15, 16) and explain the selective neural processing via phasematching oscillation between the mPFC and BLA during freezing.”

“When multiple input populations converge onto a single post-synaptic target, selective gamma synchronization can lead to selective information transmission (Akam and Kullmann, 2012; Börgers et al., 2008).”

“Brain rhythms are believed to be involved in the flow of signals around the cortex (Fries, 2005, 2009; Buschman and Miller, 2007; Saalmann et al., 2007; Pesaran et al., 2008; Wang, 2010; Akam and Kullmann, 2012; Bosman et al., 2012).”

“In fact, modeling studies investigating selective communication via coherence (Communication-through-Coherence; CTC) have focused on scenarios where information is relayed between areas (Akam and Kullmann, 2012).”

“…are di ffi cult to reconcile with the putative roles of γ -oscillations in attention and feedfor-ward inter-areal communication (Hermes et al., 2015; Akam and Kullmann, 2012; Ray and Maunsell, 2015): (1) γ - and β -activity is induced only by a subset of images and highly stimulus-dependent (Ray…”

“Communication through coherence is possible also if only synchronized rhythmic activity is present in an upstream region, as this entrains the receiver one through its fluctuating input flow (Akam & Kullmann, 2012).”

“The communication through coherence (CTC) theory predicts that the phase relationship between two populations is critical for their communication [82, 83] (also see [84, 85] for computational implementations).”

“The “communication through coherence” theory (Akam and Kullmann, 2012) refines this idea in an encoding-decoding framework, in which a top-down oscillatory signal projects to both encoding neurons with the same feature selectivity, and to the decoding network that needs to read out from them.”

“Like the communicationthrough-coherence (CTC) hypothesis [16, 26] and the frequency-division multiplexing hypothesis [12, 78], in the framework proposed here neural populations communicate dynamically along anatomical connections via coordinated rhythms.”

“The mechanisms of identical or generalized synchronization then could probably explain reasoning in such networks (see for example [44], [45]) followed by learning methods similar to any weight modification method like spike timing dependent plasticity (STDP) [46] succesfully used in artificial SNN models for pattern…”

“In addition, we analyzed whether changes in the preferred phase lag, rather than changes in phase lag stability, reflect phase-dependent modulation of functional connectivity 19,20 .”

“Neuronal filters allow neuronal systems to select certain information or enhance the communication of specific information components over others (Hutcheon & Yarom, 2000; Tsodyks et al., 2000; Izhikevich et al., 2003; Stark et al., 2022 ; Laudansky et al., 2014; Akam & Kullmann, 2012 ).”

“Indeed, modeling has revealed that population inputs containing oscillations can be selectively transmitted even in the presence of multiple distracting signals with comparable firing rates (Akam & Kullmann, 2012, 2014, 2010).”

“Computational models have highlighted strong oscillatory activity as a prerequisite for this phase-alignment 10 .”

“…and one of them is oscillatory, as long as the receiver is implemented at least as a simple spiking network with a single feed-forward interneuron layer (Akam and Kullmann, 2010), whereas filtering is much less effective when the receiver is less biologically realistic (Akam and Kullmann, 2012).”

“…and one of them is oscillatory, as long as the receiver is implemented at least as a simple spiking network with a single feed-forward interneuron layer (Akam and Kullmann, 2010), whereas filtering is much less effective when the receiver is less biologically realistic (Akam and Kullmann, 2012).”

“These observations exemplify the communication-through-coherence theory (Akam and Kullmann, 2012; Fries, 2015) and explain the selective neural processing via phase-matching oscillation between the mPFC and BLA during freezing.”

“…amplitude coupling is thought to index more stable, large-scale neural communication across distributed brain networks, thereby shaping and constraining faster, phase-based computations (Akam & Kullmann, 2012; Engel et al., 2013; Hipp et al., 2011; Lewis et al., 2016; Siegel et al., 2012).”

“This interpretation provides a putative mechanism for selective communication, in which synchronized pathways are rendered effective while non-coherent inputs, arriving out of sync with more excitable phases, are comparatively attenuated (Akam and Kullmann, 2012).”

“Neuronal filters describe the information processing building blocks where specific frequency components of the output are enhanced over others (Hutcheon and Yarom 2000; Tsodyks et al. 2000; Izhikevich et al. 2003; Stark et al. 2022; Laudansky et al. 2014; Akam and Kullmann 2012).”