Attentional resourcing : the role of cognitive complexity and cognitive load : integrating multiple- and unitary-model predictions into a two-level hierarchical framework

There are two primary attentional resource models that attempt to describe human attentional processing capabilities in multi-task situations. The unitary-resource model assumes a single store of modality-independent attentional resources which enable attentional processing for all modalities (Kahneman, 1972). It follows that performance predictions for within-modal dual-task processing (2 concurrent auditory (AA) or 2 concurrent visual (VV)) and for cross-modal dual-task processing (1 auditory concurrent with 1 visual (AV)) are similar – the two tasks compete for a share of the same resources with a consequent decrement in performance (in terms of accuracy and processing speed) irrespective of whether a cross- or within-modal task is being performed. In contrast, the multiple-resource model assumes multiple modality-specific attentional resource stores which are specific for processing in a single modality (Wickens, 1980), so it follows that this model would predict that cross-modal dual-task processing will be superior (in terms of accuracy and processing speed) to within-modal dual-task processing since in cross-modal processing the auditory and visual tasks will not compete for resources. Results from proponents of each class of model tend to support their models which could be seen as a theoretical impasse; however a review of the literature here indicates that the studies supporting each model tend to use particular methods that differ to those used in studies supporting the alternative. The proposition here is that these different methods lead to distinctly different levels of cognitive load being imposed on the individual. A model-continuum is proposed anchored at each end by these two models, with movement between these extremes being a function of the cognitive load of the task; lowest cognitive load being associated with strong multiple-resource support and high cognitive load with strong unitary-resource support. In a series of studies extraneous, intrinsic and germane cognitive load are manipulated to test the hypothesis that there will be a cross-modal advantage in conditions of low cognitive load and that as load increases the cross-modal advantage will diminish. In all the studies a dual-task target identification paradigm was used with alphabetic letters presented in rapid serial visual presentation (RSVP) and rapid auditory presentation (RAP) sequences in within-modal (AA and VV) and cross-modal (AV) conditions with objective (accuracy and reaction time) and subjective (ratings of complexity, difficulty and confidence) measures taken. In Study One no changes to the cross-modal advantage were evident across load, however subjective data suggested the cognitive load manipulation (target-distractor similarity) was ineffective. In Study Two cognitive load was manipulated by stimulus presentation rate resulting in significant cross-modal advantages under low load conditions which diminished as load increased. In Study Three the effects of motivation and task exposure were evaluated with results demonstrating that increased motivation and exposure sustained the cross-modal advantage even in high load conditions. Moreover, rated difficulty was a superior predictor of performance to rated complexity, while rated confidence was a predictor specifically in VV within-modal processing. In Study Four a hierarchical-resource model was investigated by introducing additional levels of cognitive load, and contrasting single-, dual- and tri-task performance. Results suggested that in low load conditions modality-specific resources maintain performance while in high load conditions modality-independent resources replenish these reserves. The pattern of decrement in the cross-modal advantage as a function of cognitive load suggests the Interactive Hierarchical Resource Model is a more accurate representation of human resource system than either the multiple- or unitary-resource model alone. As support for a hierarchical model means that predictions of a categorically present or absent cross-modal advantage (as in the multiple and unitary-resource models respectively) are no longer appropriate, implications with respect to performance predictions and workplace designs are discussed with an appreciation of the role of the operator’s perceptions and the load requirements of task in the degree of any cross-modal advantage