Novelty in Adaptive Information Processing

Intuitively, an adaptive information processing system should deal with unique or unusual information in a special way. For example, an unusual encounter might indicate that an organism's environment is changing, and by implication that there's a new potential for danger. Or novel information can represent a favorable change in circumstances. Either way, a system that takes special notice of infrequent stimuli is likely superior to one that is oblivious to stimulus frequency.

Recent work in the cognitive neurosciences is beginning to shed light on how the brain implements novelty or "infrequency" detection. For example, specific regions of the medial temporal lobe react to novel stimuli. Items that are familiar are processed more efficiently, as though more cognitive resources are deployed for the processing of novel items.

A 2004 article by Lavric, Pizzagalli and Forstmeier dissociates the scalp electrical currents that are associated with stimulus frequency from those associated with other cognitive processes. Two waves of electrical activity have been associated with a go/nogo task (in which subjects must make a response to frequent go stimuli, but refrain from responding to relatively infrequent nogo stimuli): a more negative wave in response to nogo stimuli relative to go stimuli (known as the N2), followed by a more positive wave in response to nogo stimuli relative to go stimuli (known as the P3).

Lavric et al. review a recent proposal associating the N2 with conflict monitoring, since source localization algorithms suggest it originates from the anterior cingulate (a region that is itself implicated in a wide variety of conflict- and error-monitoring processes). Yet other work suggests that ventral and dorsolateral prefrontal regions may contribute to the N2. What would happen if the likelihood of encountering "go" and "nogo" was equivalent - and thus equated for conflict?

To find out, Lavric et al. put a 28-electrode EEG net on the scalps of 30 subjects, each of whom completed the go/nogo task. Using the LORETA source localization algorithm, Lavric et al. were able to estimate the neural generators of scalp electrical activity.

The results showed that nogo stimuli were still associated with an N2 component, but no P3 component seemed to be present. In other words, the P3 component may be due to stimulus frequency, whereas the N2 may actually be related to inhibiting a response. Furthermore, Lavric et al. localized the N2 component to ventral and dorsolateral prefrontal cortex, again suggesting that those regions may be involved in inhibition.

However, there are at least two reasons to be cautious in accepting these conclusions.

First of all, the stimuli were matched for stimulus frequency between but not within subjects: in other words, for half the subjects, the nogo stimuli occurred 75% of the time, but the other half of subjects encountered the go stimulus 75% of the time. This is important because Lavric et al.'s analysis collapses across these two groups, without explicitly confronting whether this is warranted: the two gropus intuitively seem to be performing very different tasks. In support of this idea, reaction times were signficantly different between groups.

Secondly, inhibition is thought to be engaged when a prepotent response needs to be cancelled. In this task, half of the data points are collected from a task where there was no prepotent response. So it would be difficult to call the frontal N2 wave observed here a correlate of "inhibition" unless this effect is driven primarily by the group for whom Go stimuli occurred 75% of the time. And if that is the case, the N2 could also be a correlate of stimulus frequency detection.

Finally, other research suggests that N2 may be related to response activation, whereas the P3 is more closely related to inhibition.

In summary, Lavric et al.'s results do not unequivocally demonstrate that the N2 EEG component is sensitive to inhibition. Instead, it appears that it may be involved in conflict or error processing, or perhaps a more general monitoring role.

More like this