Cognitive Neuromodeling

High frequency oscillations (HFOs) have been empirically found in intracranial recordings and the frequency of these discrete events can help localize epileptic tissue for surgical resection. Using long-term intracranial EEG from 16 subjects, we fit Poisson and Negative Binomial mixture models that describe HFO dynamics and include the ability to switch between two discrete brain states. Oscillatory dynamics of HFO occurrences were found to be predictive of sleep state such that these model-found brain states corresponded to (1) non-REM (NREM) sleep and (2) awake and rapid eye movement (REM) sleep in two patients. In patients without expert sleep-staged data, one model-found brain state had significantly larger delta (1 - 4 Hz) power in 8/16 patients (p < .001), further suggesting that latent brain states based on HFO dynamics can predict sleep state. Parameters in each latent brain state that describe the HFO rate and clumping dynamics can be used to predict seizure onset zone (SOZ) in patients. We discuss future directions and improvements for mixture modeling of HFO dynamics. This work suggests that classification of epileptic tissue without sleep-staging can be developed using mixture modeling of HFO dynamics.

Traditionally, researchers estimate parameters of sequential sampling models (SSMs) from repeated choices across different conditions. Crucially, differences in parameters across conditions are interpreted as shifts in the underlying cognitive processes: For example, lower decision thresholds under high time pressure are interpreted as decreased cautiousness. Recent work has explored whether the parameters of SSMs can be estimated at a more detailed, single-trial level as well, to infer shifts in cognitive processes in subsequent trials. Such a more detailed window on decision-making processes has exciting applications. For example, by correlating single-trial estimates to neuroimaging data, we can relate specific brain areas to cognitive processes that may vary from trial to trial and not merely across conditions. The present work highlights some important limitations of such a powerful approach. First, we reproduce earlier work and show that single-trial estimates of SSM parameters are extremely noisy. We also show that single-trial SSM parameter estimates can be highly biased by the outcome of a choice. For example, single-trial estimates of the rate of evidence accumulation in incorrect choices are severely underestimated when compared to the generating single-trial parameter (and vice versa for correct choices). We will show how these problems can pollute the cognitive interpretation of single-trial parameters and can be exacerbated by correlations to process data. Finally, we offer a potential solution where SSMs that incorporate more information about trial-to-trial differences (e.g., stimulus or feedback properties) produce more reliable single-trial estimates.

Decision-making models based on evidence accumulation processes (the most prolific one being the drift-diffusion model – DDM) are widely used to draw inferences about latent psychological processes from chronometric data. While the observed goodness of fit in a wide range of tasks supports the model’s validity, the derived interpretations have yet to be sufficiently cross-validated with other measures that also reflect cognitive processing. To do so, we recorded electromyographic (EMG) activity along with response times (RT), and used it to decompose every RT into two components: pre-motor (PMT) and motor time (MT). These measures were mapped to the DDM’s parameters, thus allowing a test, beyond quality of fit, of the validity of the model’s assumptions and some of their usual interpretation. In two perceptual decision tasks, performed within a canonical task setting, we manipulated stimulus contrast, speed-accuracy trade-off, and response force, and assessed their effects on PMT, MT, and RT. Contrary to common assumptions, the three factors consistently affected MT. DDM parameter estimates of non-decision processes are thought to include motor execution processes, and they were globally linked to the recorded response execution MT. However, when the assumption of independence between decision and non-decision processes was not met, in the fastest trials, the link was weaker. Overall, the results show a fair concordance between model-based and EMG-based decompositions of RTs, but also establish some limits on the interpretability of decision model parameters linked to response execution.

The use of population encoding models has come to dominate human visual neuroscience, serving as a primary tool that allows researchers to infer, through indirect measurements, how cognitive states (i.e., attentional shifts, learning, adaptation, etc) change neural stimulus representations. Inverted encoding modeling is commonly used to retrieve estimates of neural population responses from neuroimaging data, but recent results suggest that the approach may have identifiability problems, because multiple mechanisms of encoding change can produce similar neural responses. Psychophysical data might be able to provide additional constraints to infer the encoding change mechanism underlying some behavior of interest. Here, we explored how well eight different mechanisms of encoding change could be differentiated by comparing the relative change between psychophysical thresholds across states. The eight types (previously proposed in the literature as mechanisms for improved task performance) included specific and nonspecific gain, specific and nonspecific tuning, specific suppression, specific suppression plus gain, and inward and outward tuning shifts. For each of the eight types (plus a homogeneous baseline), Monte Carlo simulations were used to obtain thresholds along the stimulus domain (a threshold vs stimulus function, or TvS) or along levels of external noise (a threshold vs noise function, or TvN). With the exception of specific gain and specific tuning, all studied mechanisms produced qualitatively different patterns of change in the TvN and TvS curves, suggesting that psychophysical studies can be used as a complement to inverted encoding modeling and provide strong constraints on inferences based on the latter.

Children make better decisions about perceptual information as they get older, but it is unclear how different aspects of the decision-making process change with age. Here, we used hierarchical Bayesian diffusion models to decompose performance in a perceptual task into separate processing components, testing age-related differences in model parameters and links to neural data. We collected behavioural and EEG data from 96 six- to twelve-year-olds and 20 adults completing a motion discrimination task. We used a component decomposition technique to identify two response-locked EEG components with ramping activity preceding the response in children and adults: one with activity that was maximal over centro-parietal electrodes and one that was maximal over occipital electrodes. Younger children had lower drift rates (reduced sensitivity), wider boundary separation (increased response caution) and longer non-decision times than older children and adults. Yet model comparisons suggested that the best model of children’s data included age effects only on drift rate and boundary separation (not non-decision time). Next we extracted the slope of ramping activity in our EEG components and covaried these with drift rate. The slopes of both EEG components related positively to drift rate, but the best model with EEG covariates included only the centro-parietal component. By decomposing performance into distinct components and relating them to neural markers, diffusion models have the potential to identify the reasons why children with developmental conditions perform differently to typically developing children - and to uncover processing differences which are not apparent in the response time and accuracy data alone.