Statistical learning of syllable sequences as trajectories through a perceptual similarity space.

Learning from sequential statistics is a general capacity common across many cognitive domains and species. One form of statistical learning (SL) - learning to segment "words" from continuous streams of speech syllables in which the only segmentation cue is ostensibly the transitional (or conditional) probability from one syllable to the next - has been studied in great detail. Typically, this phenomenon is modeled as the calculation of probabilities over discrete, featureless units. Here we present an alternative model, in which sequences are learned as trajectories through a similarity space. A simple recurrent network coding syllables with representations that capture the similarity relations among them correctly simulated the result of a classic SL study, as did a similar model that encoded syllables as three dimensional points in a continuous similarity space. We then used the simulations to identify a sequence of "words" that produces the reverse of the typical SL effect, i.e., part-words are predicted to be more familiar than Words. Results from two experiments with human participants are consistent with simulation results. Additional analyses identified features that drive differences in what is learned from a set of artificial languages that have the same transitional probabilities among syllables.

Stability of transcranial magnetic stimulation electroencephalogram evoked potentials in pediatric epilepsy.

Transcranial magnetic stimulation paired with electroencephalography (TMS-EEG) can measure local excitability and functional connectivity. To address trial-to-trial variability, responses to multiple TMS pulses are recorded to obtain an average TMS evoked potential (TEP). Balancing adequate data acquisition to establish stable TEPs with feasible experimental duration is critical when applying TMS-EEG to clinical populations. Here we aim to investigate the minimum number of pulses (MNP) required to achieve stable TEPs in children with epilepsy. Eighteen children with Self-Limited Epilepsy with Centrotemporal Spikes, a common epilepsy arising from the motor cortices, underwent multiple 100-pulse blocks of TMS to both motor cortices over two days. TMS was applied at 120% of resting motor threshold (rMT) up to a maximum of 100% maximum stimulator output. The average of all 100 pulses was used as a "gold-standard" TEP to which we compared "candidate" TEPs obtained by averaging subsets of pulses. We defined TEP stability as the MNP needed to achieve a concordance correlation coefficient of 80% between the candidate and "gold-standard" TEP. We additionally assessed whether experimental or clinical factors affected TEP stability. Results show that stable TEPs can be derived from fewer than 100 pulses, a number typically used for designing TMS-EEG experiments. The early segment (15-80 ms) of the TEP was less stable than the later segment (80-350 ms). Global mean field amplitude derived from all channels was less stable than local TEP derived from channels overlying the stimulated site. TEP stability did not differ depending on stimulated hemisphere, block order, or antiseizure medication use, but was greater in older children. Stimulation administered with an intensity above the rMT yielded more stable local TEPs. Studies of TMS-EEG in pediatrics have been limited by the complexity of experimental set-up and time course. This study serves as a critical starting point, demonstrating the feasibility of designing efficient TMS-EEG studies that use a relatively small number of pulses to study pediatric epilepsy and potentially other pediatric groups.

Repetitive Transcranial Magnetic Stimulation Modulates Brain Connectivity in Children with Self-limited Epilepsy with Centrotemporal Spikes.

Objective: Interictal epileptiform discharges (IEDs) alter brain connectivity in children with epilepsy; this connectivity change may be a mechanism by which epilepsy induces cognitive deficits. Here, we test whether repetitive transcranial magnetic stimulation (rTMS), a non-invasive neuromodulation technique, modulates connectivity and reduces IEDs in children with epilepsy. Methods: Nineteen children with self-limited epilepsy with centrotemporal spikes (SeLECTS) participated in a cross-over study comparing the impact of active vs. sham rTMS on IEDs and brain connectivity. SeLECTS is an epilepsy syndrome affecting the motor cortex, and prior studies show that motor cortices become pathologically hyper-connected to frontal and temporal language cortices. Using a crossover design, we compared the effect of single doses of active versus sham motor cortex rTMS. Connectivity, which was quantified by the weighted phase lag index (wPLI), was measured before and after rTMS using single pulses of TMS combined with EEG (spTMS-EEG). Analyses focused on six regions: bilateral motor cortices and bilateral inferior frontal and superior temporal regions. IEDs were counted in the five minutes before and after rTMS. Results: Active, but not sham, rTMS significantly and globally decreased wPLI connectivity between multiple regions, with the greatest reductions seen in the superior temporal region connections in the stimulated hemisphere. Additionally, there was a trend suggesting that rTMS decreases IED frequency. Interpretation: These findings underscore the potential of low-frequency rTMS to target pathologic hyperconnectivity and reduce IEDs in children with SeLECTS and potentially other pediatric epilepsy syndromes, offering a promising avenue for therapeutic intervention.