Electrical Coupling within Thalamocortical Networks Cumulatively Reduces Cortical Correlation to Sensory Inputs Authors: Austin J. Mendoza, Julie S. Haas Published: 2026 ABSTRACT Thalamocortical (TC) cells relay sensory information to the cortex as well as driving their own feedback inhibition through collateral excitation of the thalamic reticular nucleus (TRN). Inhibitory TRN cells are extensively coupled through electrical synapses. While electrical synapses are most often noted for synchronizing rhythmic forms of neuronal activity, their modulation of transient neuronal signals is less understood. Here we sought to characterize how electrical synapses embedded within a network of TRN neurons regulate the processing of ongoing sensory inputs during relay from thalamus to cortex. We constructed a thalamocortical network consisting of reciprocally connected Hodgkin–Huxley-style TC and TRN cells and one cortical output cell summing the TC activity. TRN cells were each electrically coupled to two neighboring cells, forming a ring topology. TC cells received synaptic inputs in sequence, with inputs separated by 10–50 ms, allowing us to assess the functional radius of an electrical synapse by comparing the cumulative effects of each additional TRN electrical synapse on responses within the network. Electrical synapse strength altered both TRN and TC spike response rates and latencies with each additional electrical synapse. Coupling within TRN modulated cortical integration of TC inputs by unexpectedly increasing response rates, duration, and reducing spike correlation to the input sequence that was presented to the TC layer. Thus, embedded TRN electrical synapses exert powerful influence on thalamocortical relay, highlighting the multisynaptic influences of electrically coupled cells on more complex and realistic networks of the brain. Significance Statement Electrical synapses have been widely described as synchronizers, but their role in the processing of transient information flow has received less attention. In the thalamocortical system, electrical synapses couple cells of the reticular nucleus, which inhibit and modulate sensory transmission from thalamus to cortex. We used Hodgkin–Huxley-style computational models to investigate how electrical synapses embedded in a thalamocortical network impact the neural signals passing through it. Our models reveal accumulating effects across multiple electrical synapses that ultimately and unexpectedly diminish cortical correlation to an input sequence. Our results highlight the broad reach of electrical synapses. Introduction Sensory information from the environment is communicated to the cortex by relay neurons of the thalamic nuclei. Rather than simple relay, however, it is generally agreed that neural information is transformed by thalamic circuitry. Ascending sensory information from all sensory systems, except for olfaction, targets cells of thalamic nuclei during transduction to higher brain regions (Sherman and Guillery, 2002; Jones, 2007). Thalamocortical (TC) cells within the primary sensory thalamic nuclei—the lateral geniculate nucleus for visual relay (Bickford, 2019), the ventral postmedial nucleus for touch (O'Reilly et al., 2021), and the medial geniculate body for auditory inputs (Bartlett, 2013)—receive strong “driver” inputs from sensory afferent pathways and send projections to corresponding cortical areas. Axons from TC cells also collateralize upon cells of the thalamic reticular nucleus (TRN) en route to the cortex (Scheibel and Scheibel, 1966; Lam and Sherman, 2011). The TRN is a shell of GABAergic neurons that surround the thalamus in its dorsal aspect (Pinault, 2004). Neurons of the TRN also receive excitation from corticothalamic cells of layer V and VI (Bourassa and Deschênes, 1995; Whilden et al., 2021; Carroll et al., 2022) and a wealth of modulatory input (McCormick and Wang, 1991; Pazo et al., 2013; Beierlein, 2014; Herrera et al., 2016; Aizenberg et al., 2019). In light of its potential to selectively inhibit sensory relay, TRN has been proposed to focus an attentional “searchlight” on behaviorally important information before it reaches the cortical input layer (Crick, 1984; McAlonan et al., 2008). A variety of behavioral and recording experiments have demonstrated the influence of TRN toward attentional tasks (McAlonan et al., 2000; McAlonan et al., 2006; Wimmer et al., 2015; Wells et al., 2016; Nakajima et al., 2019; Weese, 2022). Anatomical data also support potential intra-TRN and feedback connections that may implement the searchlight (Scheibel and Scheibel, 1966; Shosaku et al., 1989; Pinault et al., 1995; Pinault and Deschênes, 1998a,b). Yet the specific influences of TRN circuits on thalamocortical relay have not been well described. Communication between TRN cells is proposed as a key feature of the hypothesized spotlight function in thalamocortical circuits (Pinault and Deschênes, 1998a). Whether the GABAergic neurons of the TRN chemically synapse onto each other remains controversial. Inhibitory responses have been recorded from glutamate uncaging experiments in juvenile mice (Deleuze and Huguenard, 2006; Lam et al., 2006), but optogenetic stimulation of ChR2-expressing TRN cells revealed that direct inhibitory synapses appear to be limited to earlier development (Hou et al., 2016). Only a single direct demonstration of a mixed inhibitory and electrical connection between paired TRN cells has been reported (Landisman and Coulon, 2024). Connectivity within TRN is instead mainly thought to rely on its dense population of electrical synapses (Landisman et al., 2002). Coupling between TRN cells appears to be extensive and may couple up to 20 cells with the average coupled network size being 9 cells, as revealed by dye-coupling experiments (Lee et al., 2014). Electrical synapses can play many diverse roles within networks; the type most often described and well understood is enhancement of synchrony among coupled neurons (Lewis and Rinzel, 2003; Christie et al., 2005; Leznik and Llinás, 2005; Pfeuty et al., 2005; Mancilla et al., 2007). TRN activities are more synchronous when electrical synapses are in a stronger state (Long et al., 2004). Beyond their synchronizing influence, recent large-scale simulations of TRN, which included electrically coupled networks, have often focused on aspects of rhythmicity or sleep spindle activity (Iavarone et al., 2023; Li et al., 2024). Electrical synapses of the TRN may exert other complex effects toward processing transient inputs; models have shown that the strength of coupling can merge closely timed inputs or act to further distinguish inputs with greater differences in timing or amplitude (Pham and Haas, 2018). How electrical synapses impact the passage of transient inputs within TRN or across the brain's networks is not well understood. Previous demonstrations of transient signal processing by electrical synapses have focused on very local interactions. Coincident inputs received by two coupled cells can be enhanced through the presence of electrical synapses (Galarreta and Hestrin, 2001; Veruki and Hartveit, 2002). Similarly, temporally separated inputs to two neurons become weaker due to the current sink produced by a coupled quiescent cell (Trenholm et al., 2013; Alcami, 2018). Transmission of inhibitory afterhyperpolarizations by electrical synapses in a rhythmic context acts to desynchronize cell activity (Vervaeke et al., 2010; Hürkey et al., 2023). Coupling between cerebellar basket cells increases spiking probability in those cells through spikelet transmission, evoked within a short time window (Alcami, 2018; Hoehne et al., 2020). Similarly, electrical synapses enhance synchrony and also temporal precision in the toadfish vocal circuit (Chagnaud et al., 2021). Chemical EPSPs transmitted along a dendrite can be shunted by the presence of nearby electrical synapses (Llinás et al., 1974; Lang et al., 1996; Mendoza and Haas, 2022). These effects may also act to enhance signal-to-noise ratio in individual neurons through the combined effects described above (Ammer et al., 2022). In canonical feedforward circuits, electrical synapses between interneurons broaden the integration window of target cells and can enhance transmission efficiency in larger networks (Pham and Haas, 2019). We sought to understand how electrical synapses embedded within a network of TC and TRN neurons regulate the processing of ongoing sensory inputs during relay to cortex. Spatially extending our previous model that used a single electrical synapse between two TRN neurons to reveal modulation of the temporal dynamics between two TC cells (Pham and Haas, 2018) before they are relayed to cortex, here we constructed a coupled network of a ring of nine TRN and nine TC cells receiving a sequence of synaptic inputs, such that each subsequent TC response is modulated by the prior responses of its neighbors. We also added a single cortical integrator neuron to the model in order to examine the ultimate impacts of TRN electrical synapses upon thalamocortical relay. This architecture allowed us to examine how synapses within TRN modulated responses of the TC cells and the information that is relayed to cortex. Our simulations showed us that spiking properties and, importantly, correlation and separation of spikes in the thalamocortical relay were altered as a function of coupling strength, highlighting the complex and pervasive roles of electrical synapses in ongoing activity as it passes through networks. These results also provide a basis for understanding how plasticity of electrical synapses might affect the networks in which they are embedded. Materials and Methods Equation (1) shows the Hodgkin–Huxley equation for model cells. Models for all cells were single-point Hodgkin–Huxley models (Eq. 1), which we wrote in the Julia programing language and solved using the VCAB3 function from the DifferentialEquations.jl ODE solving suite with the third-order Adams method (Rackauckas and Nie, 2017). Simulations were run on an ASUS desktop PC with Intel i7-10700K CPU running Windows 11. Parameters for TC and TRN cells were built upon those previously reported (Destexhe et al., 1994; Traub et al., 2005; Pham and Haas, 2018) and included the following ionic currents and maximal conductances: fast transient Na+ (NaT) 60.5 mS/cm2, K+ delayed rectifier (Kd) 60 mS/cm2, K+ transient A (Kt) 5 mS/cm2, slowly inactivating K+ (K2) 0.5 mS/cm2, slow anomalous rectifier (AR) 0.025 mS/cm2, and low threshold transient Ca2+ (CaT) 0.75 mS/cm2. Reversal potentials were 50 mV for sodium, −100 mV for potassium, 125 mV for calcium, −40 mV for AR, and −75 mV for leak. Capacitance was 1 µF/cm2 with leak of 0.1 mS/cm2. Cortical output cell was a regular spiking neuron with only regular sodium and potassium currents and a leak of 0.1 mS/cm2 (Pospischil et al., 2008). Chemical synapses were modeled as double exponential decay with rise and fall time kinetics of 5 and 35 ms, respectively, with reversal potentials of 0 mV for excitatory and −100 mV for inhibitory synapses. We chose a conductance for synaptic inputs that, in the absence of other connections within the circuit, drove a four-spike burst in the TC neurons at a latency of ∼15 ms (Fig. 2B). We used excitatory chemical synapses between TC and TRN neurons strong enough to drive a three-spike burst in the TRN neurons at a latency of ∼30 ms from the input. Electrical synapses were modeled as static ohmic conductance applied to the voltage difference between the coupled TRN cells. Scaling up to a network, neuronal excitability changes as coupling is increased between neurons (Amsalem et al., 2016). Thus, for each coupling strength, we systematically scaled TRN cell leak conductance so that the neuron's input resistance matched the input resistance of the uncoupled TRN model cell. As a result of this imposition, input resistances were close to constant for each cell in the network and thereby better linked changes in spiking to changes in electrical synapse strength. Sensory inputs were simulated as exponentially decaying current pulses (peak amplitude 0.75 µA/cm2, with decay time constant 30 ms) delivered to each of the TC cells. Chemical synaptic connections between TRN cells were set at 1 µA/cm2 for excitatory synapses from TC to TRN cells and 1 µA/cm2 for all inhibitory synapses originating from TRN. For each combination of stimulus timing and electrical synapse strength, we repeated 1,500 trials; each trial lasted 1.6 s, with arrival times of spontaneous synaptic inputs drawn from a Poisson distribution at a rate of 80 Hz for excitatory and 20 Hz for inhibitory events. Coupling also increased the baseline spontaneous firing rate of the TRN neurons; therefore, we scaled the amplitude of noise to maintain all cells at a spontaneous firing rate of 6–8 Hz across all simulations. This spontaneous rate was chosen to keep spontaneous spiking low enough to distinguish input evoked responses, while also similar to awake TRN firing (Marks and Roffwarg, 1993). Subthreshold current steps were applied to all cells to bring cells close to threshold, TC and TRN cells were depolarized to −70 mV, and cortical cells were given 0.3 µA/cm2. Spike train data was extracted and saved for further analysis and visualization using MATLAB. Spike train analysis Separation of TC cell spikes were calculated as the time difference between the last spike of the first TC cell pair to receive the input and the first spike of the following cell response. Correlation was calculated from overlap of input time and spiking response time series, convolved with a falling exponential to produce similar time series to the delivered input. Latency was simply the difference between input onset and the time of the first spike in the response. Duration was the difference between the first and last spike of the response, responses consisting of only one spike were excluded from analysis. We calculated mean and standard errors of response variables across the 1,500 trials for a 100 ms time frame after input onset. Spike-triggered averages of synaptic currents were calculated from the time frame of 50 ms to 1 ms prior to the first spike of the response. Forward triggered averages were from the time frame of 1 ms to 50 ms after the first spike of the response. If no spike occurred within 100 ms following an input, the trial for that cell was excluded from calculated means and standard errors for those response characteristics. Code accessibility The code/software described in the paper is freely available online at https://github.com/jhaaslab/RingModel. The code is available as Extended Data. Download Data 1, ZIP file. Results Our general goal was to investigate the reach of electrical synapses, that is, within a network that includes coupled neurons, the radius of their functional influence. First, we measured coupling coefficients between TRN cells using the commonly used experimental paradigm of delivering a hyperpolarizing pulse to one cell and measuring voltage deflections through the rest of the network (Fig. 1A, top). For the strongest synapses, a pulse that revealed a coupling coefficient of 0.2 in the neighboring cell diminished to 0.05 at the next cell and diminished to negligible thereafter (Fig. 1B). This sharp decay of coupling coefficients suggests that electrical synapses confer only minimal influence from one neuron onto neighbors more than two synapses away, even for strong electrical synapses. We then used the same network to compare responses to identical inputs to each of the neurons (Fig. 1A, bottom), with a delay of 20 ms between inputs. In this scenario, we noted that responses of each sequential cell in the chain included more and/or faster spikes, resulting in a cumulative rather than diminishing effect on spiking in each neighboring neuron (Fig. 1C). This exercise demonstrates that coupling coefficients may not capture or predict the impact of electrical synapses within active neuronal networks. Effects of electrical synapses measured over network distance. A, Simple network of identical coupled TRN neurons. Top traces, voltage responses to a hyperpolarizing current applied to TRN1. Bottom traces, spiking responses to identical excitatory inputs (arrows) delivered to each cell. B, Coupling coefficients (cc) as a function of path length across the TRN network in A, as measured from TRN1. C, Average number of spikes in the response of each TRN cell in the network resulting from inputs shown in A. We then set out to address how electrical synapses embedded within a network of TRN neurons regulate the processing of ongoing sensory inputs relayed by a layer of thalamocortical (TC) relay neurons. To use and understand a simplified thalamic network, we provided a sequence of synaptic inputs to the TC cells of a nine-cell ring (Fig. 2A) in which each TRN neuron was electrically coupled only to its closest neighbors, rather than forming the complex syncytium of the natural TRN. In this model, distance between coupled TRN neurons is synaptic rather than physical. Because coupling introduces a current sink to a neuron that reduces input resistance (Amsalem et al., 2016), we adjusted leak conductances current in order to maintain TRN input resistances at a constant value as we varied coupling across simulations. All cells were provided with spontaneous input such that they fired spontaneously at 6–8 Hz (Fig. 2B; Marks and Roffwarg, 1993); this choice rendered them more sensitive to the smaller synaptic inputs we provided, compared with neurons at rest that would require larger inputs. Synaptic inputs were delivered to TC cells at a fixed time difference (Δtin), between 10 and 50 ms; we chose this set of inputs to represent a spectrum of thalamic inputs ranging from nearly identical to less similar (e.g., whiskers moving across a surface; similarly timed or frequencies of sound). These choices and input sequences gave us a basis for evaluating the impact of electrical synapses between TRN neurons, whose effects were exerted through the inhibitory synapses from each TRN cell to its paired TC cell, on thalamic relay of synaptic inputs. Coupling within the TRN layer modulates the latency and duration of TRN and TC spike trains. A, Schematic of the ring network and inputs. B, Example rasters of TRN and TC spike trains. C, Representative trace showing measurements of spiking properties resulting from an input delivered to one TC cell. Scale bars: 25 mV, 20 ms. D, Average spiking latency of TRN cells in the network with increasing Gc. Black lines represent latency in an uncoupled network. Left to right panels, increasing Δtin. E, Average spiking latency of TC cells with increasing Gc within the TRN cell network. Black lines represent values in an uncoupled network. Left to right panels, increasing Δtin. F, Average spiking duration of TRN cells in the network with increasing Gc value. Black lines represent duration in an uncoupled network. Left to right panels, increasing Δtin. G, Average spiking duration of TC cells with increasing Gc within the TRN cell network. Black lines represent duration in an uncoupled control network. Left to right panels, increasing Δtin. H, Average gap junctional current (IGJ) preceding spikes (t = 0) in TRN cell 1 (left), TRN cell 5 (middle), TRN cell 9 (right) for Δtin = 10 ms across values of Gc. I, As in H, for Δtin = 50 ms. J, Average inhibitory current (IGABA) following a spike (t = 0) received by TC cell 1 (left), TC cell 5 (middle), TC cell 9 (right), for Δtin = 10 ms across values of Gc. K, As in J, for Δtin = 50 ms. To understand the impacts of electrical synapses within the TRN layer on TC spiking, we first measured the latency and duration of spiking responses in the TRN cells as a function of electrical synapse strength and input timing (Fig. 2C). In the TRN neurons, increases in electrical synapse strength increased the latency of TRN spiking (Fig. 2D), consistent with relative increases in the current sink in IGJ imposed by stronger gap junctions to inactive neighbors. This effect was proportional to electrical synapse strength and was largest for TRN1 (Fig. 2H, left). Increased latency from the sink was countered by gap junction-mediated spread of input-driven excitation that accumulated in later TRN neurons (Fig. 2H, right). These results illustrate that by spreading either excitation of spiking or inhibition of a sink, electrical synapses bidirectionally amplify ongoing activity or its absence. Latency changes in TRN were largest for more closely timed inputs, where excitation from one input overlapped with arrival of the next input; for more separated inputs, the sink effect was larger (Fig. 2I). Duration of TRN spike bouts was also modulated by electrical synapses, with consistently longer duration proportional to coupling strength, as expected, from shared excitation between TRN neurons for shorter Δtin. TRN duration diminished with increased Δtin (Fig. 2F), as shared excitation decayed more between bouts of spiking. Overall, effects of electrical synapses within TRN cells were strongest for the neurons closest to the initial input and diminish for the more distant cells, as expected. Next, we turned to the TC layer. Because inputs arrived directly to TC cells first, latency of spiking was only minimally later in TC cells with greater TRN coupling strength (Fig. 2E). Duration of TC spiking (Fig. 2G) was strongly controlled by the inhibition delivered from TRN cells during a spiking bout (IGABA; Fig. 2J), which was in turn modulated by the strength of electrical synapses and shared excitation between TRN neurons. As a result, TC spike trains were truncated by up to 5 ms when both coupling was high between TRN cells and Δtin was small (Fig. 2G, left). We also noted a cumulative increase in the baseline tonic (t = 0) inhibitory currents from TC1 (0.015 µA) to TC9 (0.078 µA) for Δtin = 10 ms, resulting from cumulative shared activity across TRN neurons. For larger Δtin, shorter duration of TRN bouts, domination with the TRN layer of the sink from gap junctions, and less overlap between inputs resulted in less inhibition delivered to TC neurons as electrical synapse strength increased (Fig. 2K), allowing the TC duration to increase progressively. Electrical synapses in the TRN layer also modulated the correlations of TC and TRN neurons to the input. We measured correlation between the input times and spike trains in the coupled network, each convolved by a falling exponential with time constant 5 ms. TRN cells were least correlated to inputs for closely timed inputs and strong electrical synapses (Fig. 3A, left), diminishing by 50%, as a result of the GJ-driven changes in latency and duration of their spike trains described above. A cumulative effect of electrical synapses can be seen across TRN cells, producing a gradual increase in correlation from the first to last TRN to receive the input. Correlation of TC cells to their inputs was also modulated, although much less strongly, by GJs between the TRN neurons, and again in a cumulative manner across TC cells (Fig. 3B). These effects were driven mainly by changes in duration of TC spiking, which differed most greatly at the extremes of Δtin and produced the lowest correlations at those values as well. Electrical coupling decorrelates TRN and TC trains from sensory inputs. A, Average correlation between input trains and TRN spiking for increasing Gc value. Black line represents correlation for an uncoupled network. Left to right panels, increasing Δtin. B, Average correlation between input trains and TC spiking for increasing Gc within the TRN layer. Black line represents separation correlation in an uncoupled network. Left to right panels, increasing Δtin. As a first step at understanding how cortical reception of thalamocortical relay is modulated by TRN electrical synapses, we measured the separation and overlap between TC spike train bouts (Fig. 4A; Pham and Haas, 2018). Separation, measured in ms, is the time gap between spike bouts and provides an indication of whether a postsynaptic neuron might be expected to discriminate between two bouts of spiking. Greater separation between TC spike trains potentially underlies discrimination by a cortical neuron that is integrating TC inputs. As a result of inhibition from the TRN spike trains, coupling strength altered the separation values of each pair of TC spike trains. When Δtin was small, TC spikes had significant overlap (separation < 0) in the uncoupled networks; increasing electrical synapse strength produced increased separation and reduced overlap (Fig. 4B, left). At longer Δtin, where TC spikes were already positively separated, the overall separation in TC spike trains was reduced by greater electrical coupling (Fig. 4B, right). In both cases, the effects increased across pair number as more inputs were delivered to the TC layer, again demonstrating a cumulative effect due to network coupling (Fig. 4C). The magnitude of separation changes was considerable at the timescale of spike timing, as the effect of coupling produced up to 10 ms difference from control spike separations (Fig. 4D). In summary, the decorrelation of TC cell responses and thus changes in how separated responses were from control are produced by both latency and duration changes dependent on the coupling within the network. The consequences of these temporal changes predict substantial changes in how the cortex might integrate and process the arrival of these thalamocortical inputs. Coupling within TRN layer modulates separation of TC spike trains. A, Representative traces from the first three TC cells for a control trial (no coupling between TRN cells). Left traces show separation with a small Δtin, and right traces show separation with a large Δtin, with separation notated. Horizontal dotted lines indicate the start and end of each spiking bout used to measure separation. B, As in A, for a trial with high coupling strength between TRN cells. C, Average pairwise spike separations of TC cells in the network with increasing values of uniform Gc between TRN cells. Black lines represent separation in an uncoupled network. Left to right, increasing Δtin. D, Average pairwise changes in spike separation of TC cells, plotted for increasing Gc within the TRN cell network, relative to the control response of TC to input alone (black lines). Panels show results for increasing Δtin. Finally, we added a single simple integrator neuron to sum and read out the output of thalamocortical relay from our ring network (Fig. 5A). Because spontaneous TC spikes passed through TRN gap junctions could confound or mask the effects of electrical synapses passing input-driven TC spikes within the TRN layer on the ultimate cortical integration of those inputs, we ran these simulations in the absence of noise. Instead, to produce variation across trials, we delivered inputs with 5% variation in the input arrival times. We chose an excitatory synapse strength for TC inputs to the cortical neuron that produced a single spike in the cortical neuron for each TC input (Fig. 5B, bottom). Cortical spike times were nearly regular (CV of spike trains = 20–30%) for no or low TRN coupling. Surprisingly, we observed that increases in coupling within the TRN layer produced more spikes, irregular spikes within trains, and longer spike trains in the cortex cell across all Δtin values (Fig. 5B). Interspike interval (ISI) distributions for cortical spikes were tightly restricted around the Δtin value in the uncoupled and low-coupling ranges (Fig. 5C, bottom). With increased coupling strength within the TRN, ISI distributions in the cortical neuron broadened substantially (Fig. 5C, top). To better visualize this effect, we plotted the gain of the cortical ISI distributions in coupled cases relative to the ISI distributions for uncoupled TRN (Fig. 5D). These plots revealed a reduction in ISIs around the value of Δtin (dotted line) and a gain in shorter ISIs as coupling strength increased, indicative of the shift from regular spiking toward more irregular firing. Consistent with the increase in shorter ISIs, the average spiking output of the cortex cell increased for higher coupling within TRN (Fig. 5E). We quantified the effect on irregular firing by measuring the coefficient of variation (CV) of the ISI distribution. Spike trains in cortex were increased in CV to 60% at higher coupling values and across Δtin (Fig. 5F). Lastly, we asked if these spiking changes modulated the correlation of cortex spikes to the TC input train. Correlation in the cortex spikes to the input train was high (70–90%) for uncoupled and weakly coupled TRN and decreased (to 40–50%) as a result of increased coupling (Fig. 5G). We conclude that electrical synapses within the TRN have diverse impacts on TRN spiking, which accumulated and propagated to modulate TC cell responses, and ultimately, cortical responses to thalamocortical inputs. Impact of coupling within TRN on TC-relayed cortical spiking. A, We added a cortical neuron to integrate TC relay from the ring network. B, Cortical spike trains for control (no coupling between TRN cells), moderate (blue), and high (yellow) coupling between TRN cells within the network, for Δtin = 10 ms (left) and Δtin = 50 ms (right). C, ISI distributions for the cortex cell for control (no coupling between TRN cells), moderate (blue), and high (yellow) coupling between TRN cells of the network. Mean ISI distributions for uncoupled TRN are overlayed in red. Left to right, increasing Δtin. D, Gain of the ISI distributions, versus time and across coupling values. E, Average number of spikes in the cortex cell, versus Δtin and across coupling values. F, Average coefficient of variation of the ISI distributions in C, versus Δtin and across coupling values. G, Average correlation between cortex spike times and input sequence times, versus Δtin and across coupling values. Discussion While the role of electrical synapses in synchronizing rhythmic activity is well established, their contributions to the neuronal processing of transient inputs across networks remains less well characterized. Our computational investigation demonstrates that embedded electrical synapses have an unexpectedly decorrelating and potentially strong impact on thalamocortical relay that arises as the modulation of electrical synapses between TRN neurons propagates and accumulates through the network. Electrical synapses between TRN cells modulate latency and especially duration of TRN spike trains; these changes are passed into the TC cells through inhibitory synapses and impact the duration and correlation of TC spiking. The effects of each electrical synapse within the chain accumulate with each cell and result in dispersed and decorrelated responses in the cortical cell. These effects illustrate that TRN coupling influences activity through a large extended network, with effects on spiking output that would not be predicted from the passive measurements of each electrical synapse alone, as coupling between pairs alone would not predict the effects we see in a network context and drops to indistinguishable levels when measured by the traditional hyperpolarizing current steps across multiple synapses. The TRN is hypothesized to focus a “searchlight” for information transmitted to cortex by thalamus (Crick, 1984; McAlonan et al., 2008), but the details of such a process remain underdescribed. Our results indicate that thalamocortical relay may be strongly modulated, and information transformed, by electrical coupling within TRN, as our results suggest ongoing activity is less faithfully transmitted when TRN cells are strongly coupled. We hypothesize that coupling might be modulated within small subnetworks of the TRN by activity or neuromodulators to selectively impact thalamocortical transmission. Numerous mechanisms of plasticity have been described in coupled systems (Lasater and Dowling, 1985; Pereda and Faber, 1996; Landisman and Connors, 2005; Haas et al., 2011; Mathy et al., 2014; Wang et al., 2015). Coupling across TRN could be temporarily upregulated across a sensory sector, or globally, in order to disrupt attention to the sensory surround, for instance. Alternatively, relay could be improved by weakened coupling in the relay channel for a specific object while increased coupling might disperse and interrupt relay in a neighboring channel, resulting in increased cortical reception of and attention to of the first channel. Many such schemes are possible. Though we did not include plasticity dynamics in our model, the effects of modifying synapse strengths within the network can be extrapolated from the synapse strengths we sampled in our network model to provide specific predictions. TRN cells couple small clusters of nearby cells with direct electrical synapses; our model here considers a specific topology of connections in order to address the extent of electrical synapse influence over spike times. Effects of gap junctions on transient inputs, which note excitatory effects for closely timed inputs, would also propagate across several cells in a network context and have considerable impacts on information processing (Galarreta and Hestrin, 2001; Vervaeke et al., 2010; Trenholm et al., 2013; Pham and Haas, 2018; Hoehne et al., 2020). Our simulations provide basic investigations into larger electrical synaptic embedded networks at TRN, which can form a basis of comparison for future investigations. In particular, whether these results can be reproduced in a more naturalistic topology in which connections are randomly dispersed among nearby cells that form a syncytium remains to be determined. In live tissue, electrical synapses are stronger and more numerous for cells with closer intersomatic distances (Long et al., 2004). The results here, where cortical integration is influenced by alterations in TC firing resulting from TRN coupling, may vary in vivo. We expect that impacts on cortical integration would be strongest for nearby TC cells if the TRN cells they receive inputs from are highly coupled. TC cells may receive input from multiple coupled TRN neurons, possibly even amplifying these effects. Many refinements are possible for our network model that could impact its predictions. Our thalamic cell models were detailed in order to include the important bursting behaviors in those cell types (McCormick and Huguenard, 1992; Destexhe et al., 1996; Destexhe et al., 1998), but we opted for a simplistic cortical output neuron so that changes we observed in responses were due to the electrical synapse-mediated changes in the relay alone. We expect that intrinsic properties of cortical cells could be another important source of modulation in cortical response that would interact with the modulation by TRN synapses. Thus, further study could include more detailed neuron models for layer 4 and include specifics such as spike adaptation due to M-currents or dendritic morphology and dynamics. Further, deep-layer cortical neurons provide an important source of input to TRN (Gentet and Ulrich, 2004; Whilden et al., 2021; Carroll et al., 2022; Hádinger et al., 2023; Sokhadze et al., 2025), and the impacts of that connection combined with electrical synapses within TRN should also be expected to modulate thalamocortical transmission. Also absent from our models are lateral inhibitory connections between nearby TRN cells (Deleuze and Huguenard, 2006; Lam et al., 2006; Landisman and Coulon, 2024). Corticothalamic feedback onto both TC and TRN neurons, although delayed, could be another avenue of investigation for future experiments. Another key thalamic synapse proposed as a mechanism of the searchlight is open-loop divergent inhibition from TRN to TC cells (Crabtree et al., 1998; Crabtree and Isaac, 2002). TRN has recently been shown to comprise distinct cell types which differentially target thalamic relays of first and higher-order information (Clemente-Perez et al., 2017; Martinez-Garcia et al., 2020). Our results here only consider one homogenous source of TC relay and do not include the influence of these cell types and the potential sensory modulation that higher-order sources may contribute (Bu et al., 2025). Electrical synapses have been shown to couple TRN cells across cell types (Vaughn et al., 2024); additionally, somatostatin-expressing TRN may provide lateral inhibition to the first-order parvalbumin-expressing TRN cell type (Rolón-Martínez et al., 2026). In all, our model can provide important predictions for the diverse connections both within and between thalamic sources and their impact on precise sensory relay to the cortex.