Skip to main content
Log in

Neural Habituation Enhances Novelty Detection: an EEG Study of Rapidly Presented Words

Computational Brain & Behavior Aims and scope Submit manuscript

Cite this article


Huber and O'Reilly (Cognitive Science, 27(3), 403–430, 2003) proposed that neural habituation aids perceptual processing, separating neural responses to currently viewed objects from recently viewed objects. However, synaptic depression has costs, producing repetition deficits. Prior work confirmed the transition from repetition benefits to deficits with increasing duration of a prime object, but the prediction of enhanced novelty detection was not tested. The current study examined this prediction with a same/different word priming task, using support vector machine (SVM) classification of EEG data, event-related potential (ERP) analyses focused on the N400, and dynamic neural network simulations fit to behavioral data to provide a priori predictions of the ERP effects. Subjects made same/different judgements to a response word in relation to an immediately preceding brief target word; prime durations were short (50 ms) or long (400 ms), and long durations decreased P100/N170 potentials to the response word, suggesting that this manipulation increased habituation. Following long duration primes, correct “different” judgments of primed response words increased, evidencing enhanced novelty detection. An SVM classifier predicted trial-by-trial behavior with 66.34% accuracy on held-out data, with greatest predictive power at a time pattern consistent with the N400. The habituation model was augmented with a maintained semantics layer (i.e., working memory) to generate behavior and N400 predictions. A second experiment used response-locked ERPs, confirming the model’s assumption that residual activation in working memory is the basis of novelty decisions. These results support the theory that neural habituation enhances novelty detection, and the model assumption that the N400 reflects updating of semantic information in working memory.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Institutional subscriptions

Fig. 1
Fig. 2
Fig. 3
Fig. 4
Fig. 5
Fig. 6
Fig. 7
Fig. 8
Fig. 9
Fig. 10

Similar content being viewed by others

Data Availability Statement

The computational model code used for generating model predictions and the EEG datasets analyzed during the current study are available in the following OSF repository:


Download references


We thank Anushree Mehta for her work on an earlier version of the SVM classifier, and Christoph Weidemann for his suggestions towards improving our classifier analyses.

Author information

Authors and Affiliations


Corresponding author

Correspondence to Len P. L. Jacob.

Additional information

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Electronic Supplementary Material

Supplementary Fig. 1

Output and resources for the response word node within the maintained semantic layer of the model. In the model, higher output will consume resources, and decreased resources will reduce output (see “Habituation model” subsection within methods section for details and equations). In this figure, output and resources of the node corresponding to the response word (bolded in each subplot) is shown across a full trial. Residual activations, the lowest output value following response word presentation, are also marked. Lower residual activations cause the model to predict a “different” response, while higher residual activations cause it to predict a “same” response. Accuracy for each of the 8 conditions is determined from the value indicated by the arrow, as compared to a criterion. We consider each condition in turn, working up from the bottom. In the case of the A-A-B conditions, there is no residual activation for B, and so accuracy is good and unchanged by prime duration. In the case of the A-B-B condition, residual activation for B indicates a correct answer of “same”. Residual activate for B increases with increasing prime duration because as habituation for A increases, it does not compete as much with B in the perceptual layers of the model (i.e., a better response to the briefly flashed target word B). In the case of the A-B-A conditions, residual activation from the prime (word A) incorrectly indicates a “same” response. However, for a longer duration prime, word A is habituated, and so there is less residual activation for A, which improves accuracy (enhanced novelty detection). Finally, in the case of A-A-A, residual activation for A correctly indicates a “same” response. However, habituation for word A weakens this residual activation, and reduces accuracy with increasing prime duration (i.e., repetition blindness leads to worse performance). (PNG 92 kb)

High Resolution Image (EPS 2354 kb)

Supplementary Fig. 2

Layer specific activation profiles. The activation profiles of multiple layers were combined to generate N400 and P100/N170 ERP predictions; the above shows the activation of the individual layers. These waveforms were obtained by summing the output of all nodes within a layer at each unit of simulated time. See “Habituation model” subsection within methods section for information on how parameters were obtained. (PNG 102 kb)

High Resolution Image (EPS 2685 kb)

Supplementary Fig. 3

Equivalent model fitting results as compared to the model fits shown in Fig. 4, except that observed data have been time-reversed by labeling the actual 400 ms prime duration conditions as being the 50 ms prime duration conditions and the actual 50 ms prime duration conditions as being the 400 ms prime duration conditions. As in Figs. 4, 5 parameters were optimized in attempt to explain the 8 conditions. Despite having the same 5 free parameters to capture these time-reversed data, the model completely failed to account for the results. The best-fitting χ2 for these time-reversed data was 548.4, which can be contrasted with a χ2 of 40.4 for the results show in Fig. 4. In this case, the best that the model could do was to minimize the role of prime duration to capture the overall average accuracy collapsed across the conditions. This highlights that this is a dynamic systems model, rather than a measurement model. The model is greatly constrained in its predictions for changes across manipulations of duration. More specifically, as prime duration increases, the model necessarily predicts that habituation will increase. (PNG 58 kb)

High Resolution Image (EPS 1173 kb)

Rights and permissions

Reprints and Permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Jacob, L.P.L., Huber, D.E. Neural Habituation Enhances Novelty Detection: an EEG Study of Rapidly Presented Words. Comput Brain Behav 3, 208–227 (2020).

Download citation

  • Published:

  • Issue Date:

  • DOI: