Artificial Neural Network Language Models Predict Human Brain Responses to Language Even After a Developmentally Realistic Amount of Training.

Artificial neural networks have emerged as computationally plausible models of human language processing. A major criticism of these models is that the amount of training data they receive far exceeds that of humans during language learning. Here, we use two complementary approaches to ask how the models' ability to capture human fMRI responses to sentences is affected by the amount of training data. First, we evaluate GPT-2 models trained on 1 million, 10 million, 100 million, or 1 billion words against an fMRI benchmark. We consider the 100-million-word model to be developmentally plausible in terms of the amount of training data given that this amount is similar to what children are estimated to be exposed to during the first 10 years of life. Second, we test the performance of a GPT-2 model trained on a 9-billion-token dataset to reach state-of-the-art next-word prediction performance on the human benchmark at different stages during training. Across both approaches, we find that (i) the models trained on a developmentally plausible amount of data already achieve near-maximal performance in capturing fMRI responses to sentences. Further, (ii) lower perplexity-a measure of next-word prediction performance-is associated with stronger alignment with human data, suggesting that models that have received enough training to achieve sufficiently high next-word prediction performance also acquire representations of sentences that are predictive of human fMRI responses. In tandem, these findings establish that although some training is necessary for the models' predictive ability, a developmentally realistic amount of training (∼100 million words) may suffice.

The neural architecture of language: Integrative modeling converges on predictive processing.

The neuroscience of perception has recently been revolutionized with an integrative modeling approach in which computation, brain function, and behavior are linked across many datasets and many computational models. By revealing trends across models, this approach yields novel insights into cognitive and neural mechanisms in the target domain. We here present a systematic study taking this approach to higher-level cognition: human language processing, our species' signature cognitive skill. We find that the most powerful "transformer" models predict nearly 100% of explainable variance in neural responses to sentences and generalize across different datasets and imaging modalities (functional MRI and electrocorticography). Models' neural fits ("brain score") and fits to behavioral responses are both strongly correlated with model accuracy on the next-word prediction task (but not other language tasks). Model architecture appears to substantially contribute to neural fit. These results provide computationally explicit evidence that predictive processing fundamentally shapes the language comprehension mechanisms in the human brain.

An Open Resource for Non-human Primate Optogenetics.

Optogenetics has revolutionized neuroscience in small laboratory animals, but its effect on animal models more closely related to humans, such as non-human primates (NHPs), has been mixed. To make evidence-based decisions in primate optogenetics, the scientific community would benefit from a centralized database listing all attempts, successful and unsuccessful, of using optogenetics in the primate brain. We contacted members of the community to ask for their contributions to an open science initiative. As of this writing, 45 laboratories around the world contributed more than 1,000 injection experiments, including precise details regarding their methods and outcomes. Of those entries, more than half had not been published. The resource is free for everyone to consult and contribute to on the Open Science Framework website. Here we review some of the insights from this initial release of the database and discuss methodological considerations to improve the success of optogenetic experiments in NHPs.

Dissociating effects of error size, training duration, and amount of adaptation on the ability to retain motor memories.

Extensive computational and neurobiological work has focused on how the training schedule, i.e., the duration and rate at which an environmental disturbance is presented, shapes the formation of motor memories. If long-lasting benefits are to be derived from motor training, however, retention of the performance improvements gained during practice is essential. Thus a better understanding of mechanisms that promote retention could lead to the design of more effective training procedures. The few studies that have investigated how retention depends on the training schedule have suggested that the gradual exposure of a perturbation leads to improved retention of motor memory compared with an abrupt exposure. However, several of these previous studies showed small effects, and although some controlled the training duration and others the level of learning, none have controlled both. In the present study we disambiguated both of these effects from exposure rate by systematically varying the duration of training, type of trained dynamics, and exposure rate for these dynamics in human force-field adaptation. After controlling for both training duration and the amount of learning, we found essentially identical retention when comparing gradual and abrupt training for two different types of force-field dynamics. By contrast, we found that retention was markedly higher for long-duration compared with short-duration training for both types of dynamics. These results demonstrate that the duration of training has a far greater effect on the retention of motor memory than the exposure rate during training. We show that a multirate learning model provides a computational mechanism for these findings.NEW & NOTEWORTHY Previous studies have suggested that a gradual, incremental introduction of a novel environment is helpful for improving retention. However, we used experimental and computational approaches to demonstrate that previously reported improvements in retention associated with gradual introductions fail to persist when other factors, including the duration of training and the degree of initial learning, are accounted for.

The 24-h savings of adaptation to novel movement dynamics initially reflects the recall of previous performance.

Humans rapidly adapt reaching movements in response to perturbations (e.g., manipulations of movement dynamics or visual feedback). Following a break, when reexposed to the same perturbation, subjects demonstrate savings, a faster learning rate compared with the time course of initial training. Although this has been well studied, there are open questions on the extent early savings reflects the rapid recall of previous performance. To address this question, we examined how the properties of initial training (duration and final adaptive state) influence initial single-trial adaptation to force-field perturbations when training sessions were separated by 24 h. There were two main groups that were distinct based on the presence or absence of a washout period at the end of day 1 (with washout vs. without washout). We also varied the training duration on day 1 (15, 30, 90, or 160 training trials), resulting in 8 subgroups of subjects. We show that single-trial adaptation on day 2 scaled with training duration, even for similar asymptotic levels of learning on day 1 of training. Interestingly, the temporal force profile following the first perturbation on day 2 matched that at the end of day 1 for the longest training duration group that did not complete the washout. This correspondence persisted but was significantly lower for shorter training durations and the washout subject groups. Collectively, the results suggest that the adaptation observed very early in reexposure results from the rapid recall of the previously learned motor recalibration but is highly dependent on the initial training duration and final adaptive state.NEW & NOTEWORTHY The extent initial readaptation reflects the recall of previous motor performance is largely unknown. We examined early single-trial force-field adaptation on the second day of training and distinguished initial retention from recall. We found that the single-trial adaptation following the 24-h break matched that at the end of the first day, but this recall was modified by the training duration and final level of learning on the first day of training.

Flexible Sensorimotor Computations through Rapid Reconfiguration of Cortical Dynamics.

Neural mechanisms that support flexible sensorimotor computations are not well understood. In a dynamical system whose state is determined by interactions among neurons, computations can be rapidly reconfigured by controlling the system's inputs and initial conditions. To investigate whether the brain employs such control mechanisms, we recorded from the dorsomedial frontal cortex of monkeys trained to measure and produce time intervals in two sensorimotor contexts. The geometry of neural trajectories during the production epoch was consistent with a mechanism wherein the measured interval and sensorimotor context exerted control over cortical dynamics by adjusting the system's initial condition and input, respectively. These adjustments, in turn, set the speed at which activity evolved in the production epoch, allowing the animal to flexibly produce different time intervals. These results provide evidence that the language of dynamical systems can be used to parsimoniously link brain activity to sensorimotor computations.

Flexible timing by temporal scaling of cortical responses.

Musicians can perform at different tempos, speakers can control the cadence of their speech, and children can flexibly vary their temporal expectations of events. To understand the neural basis of such flexibility, we recorded from the medial frontal cortex of nonhuman primates trained to produce different time intervals with different effectors. Neural responses were heterogeneous, nonlinear, and complex, and they exhibited a remarkable form of temporal invariance: firing rate profiles were temporally scaled to match the produced intervals. Recording from downstream neurons in the caudate and from thalamic neurons projecting to the medial frontal cortex indicated that this phenomenon originates within cortical networks. Recurrent neural network models trained to perform the task revealed that temporal scaling emerges from nonlinearities in the network and that the degree of scaling is controlled by the strength of external input. These findings demonstrate a simple and general mechanism for conferring temporal flexibility upon sensorimotor and cognitive functions.

The decay of motor adaptation to novel movement dynamics reveals an asymmetry in the stability of motion state-dependent learning.

Motor adaptation paradigms provide a quantitative method to study short-term modification of motor commands. Despite the growing understanding of the role motion states (e.g., velocity) play in this form of motor learning, there is little information on the relative stability of memories based on these movement characteristics, especially in comparison to the initial adaptation. Here, we trained subjects to make reaching movements perturbed by force patterns dependent upon either limb position or velocity. Following training, subjects were exposed to a series of error-clamp trials to measure the temporal characteristics of the feedforward motor output during the decay of learning. The compensatory force patterns were largely based on the perturbation kinematic (e.g., velocity), but also showed a small contribution from the other motion kinematic (e.g., position). However, the velocity contribution in response to the position-based perturbation decayed at a slower rate than the position contribution to velocity-based training, suggesting a difference in stability. Next, we modified a previous model of motor adaptation to reflect this difference and simulated the behavior for different learning goals. We were interested in the stability of learning when the perturbations were based on different combinations of limb position or velocity that subsequently resulted in biased amounts of motion-based learning. We trained additional subjects on these combined motion-state perturbations and confirmed the predictions of the model. Specifically, we show that (1) there is a significant separation between the observed gain-space trajectories for the learning and decay of adaptation and (2) for combined motion-state perturbations, the gain associated to changes in limb position decayed at a faster rate than the velocity-dependent gain, even when the position-dependent gain at the end of training was significantly greater. Collectively, these results suggest that the state-dependent adaptation associated with movement velocity is relatively more stable than that based on position.