Neural correlates of irrelevant speech

Abstract

The irrelevant speech effect refers to a reduction in the immediate serial recall of lists of presented items when irrelevant auditory material is presented together with the items to be memorized. The ability to ignore irrelevant distractors is vital in increasingly demanding work environments and the capability to keep information in and out of mind is central in human information processing. Hence, the scope of the study of irrelevant speech effects ranges from implications for theories of attention and working memory, to practical consequences of irrelevant speech in work environments or command and control centers. Whereas the detrimental effect of irrelevant speech is well established in the behavioral literature, much less is known about its neurophysiological correlates. As evident from the different theoretical accounts of the irrelevant speech effect, there is no general agreement on which representational level is affected by the interference. The hypotheses cover different stages in the information flow, ranging across perception, attention, and memory.

The overall purpose of the present research is to provide an understanding of how the human brain functions when speech interferes with the processing of visually presented material in general, and, more specifically, to explore some possible propositions as to the representational level (e.g. perceptual, attentional, or mnemonic) of the irrelevant speech effect.

In the present thesis, behavioral experiments combined with PET methodology were used to investigate the neurophysiological mechanisms behind the irrelevant speech effect. In a network approach using structural equations modeling, we also investigated the functional and effective connectivity between different regions of the brain. In addition, we employed a new behavioral task paradigm known as the serial recognition task. This task has several potential benefits compared to a standard serial recall task. Most importantly, serial recognition requires primarily that the serial order is maintained in short-term memory while minimizing the need for a representation of the content of the presented items as well as output systems.

In an initial study we used an exploratory, low level of working memory load. We found that the irrelevant speech effect correlated with a decrease in regional blood flow in several areas previously associated with verbal working memory. These areas were located in the bilateral superior temporal and inferior/middle frontal cortices extending into Broca's area on the left, as well as in the left inferior parietal cortex. The results suggest that areas associated with verbal working memory were broadly suppressed during irrelevant speech.

In paper II, the working memory load was increased and corresponded to a conventional behavioral study of the irrelevant speech effect. The decreases in activity that were observed in the first low load study were not replicated in this study. However, in the reversed contrast, an increase in bilateral dorsolateral prefrontal activity was observed. This area is frequently associated with executive aspects of working memory and attentional control mechanisms.

In paper III we investigated the functional and effective connectivity in the dataset from study I. Initial analyses revealed an interaction between the level of irrelevant speech and the functional connectivity between the regions related to verbal working memory and the left medial temporal lobe. We investigated network interactions between the working memory network and the medial temporal lobe, including different functional sub-networks. No meaningful functional sub-network differed significantly between conditions except for the connections related to the medial temporal lobe. Based on these observations, we propose that the exposure of irrelevant speech results in a switch from a relatively pure processing of phonological representations to an engagement in processing of episodic representations and that this is indicated by a stronger interaction between the verbal working memory network and the medial temporal lobe.

In the fourth and final study we observed a significant irrelevant speech effect of changing versus steady-state auditory items in a serial recognition task. This finding lends support for and extends previous empirical findings, suggesting that irrelevant speech has the potential to interfere with the coding of the order of the items to be memorized.

Taken together, we suggest that these studies provide findings as to the functional neuroanatomical nature of the irrelevant speech effect. Firstly, there is no single neurophysiological locus of the irrelevant speech effect. The results of the present studies do not exclude any of the hypothesized levels at which the effect occurs, rather it appears that all levels are affected.

Secondly, functional neuroirnaging methods can provide additional information about the mechanisms underlying the irrelevant speech effect that is difficult, if not impossible, to obtain with behavioral methods alone.

Finally, within the paired subtraction paradigm, changes in rCBF take place in areas that are inherently activated by the verbal working memory task itself. However, as reflected in the network analysis, dynamic changes between these areas and the medial temporal lobe, suggest that new areas are recruited according to task demands.