Routes to short-term memory indexing: lessons from deaf native users of American Sign Language

Abstract

Models of working memory (WM) have been instrumental in understanding foundational cognitive processes and sources of individual differences. However, current models cannot conclusively explain the consistent group differences between deaf signers and hearing speakers on a number of short-term memory (STM) tasks. Here we take the perspective that these results are not due to a temporal order-processing deficit in deaf individuals, but rather reflect different biases in how different types of memory cues are used to do a given task. We further argue that the main driving force behind the shifts in relative biasing is a consequence of language modality (sign vs. speech) and the processing they afford, and not deafness, per se.

Figure 1

The multiple component working memory model adapted from Baddeley (2000) which accounts for relative group biases. Incoming information first activates long-term memory (LTM) representations. This system permits the accumulation of long-term knowledge. In contrast, the short-term memory buffers represent dedicated temporary workspace, all under the control of the central executive (not shown here). The episodic buffer binds information from multiple codes and provide the richest input to the central executive, whereas the two slave systems are highly specialized for phonological versus visuo-spatial information respectively. More over, the episodic system uniquely labels events with respect to their time and place of occurrence allowing individuation of “episode” or temporary memory trace that uniquely refer to one event in the world. In contrast, temporal information in the phonological loop is conveyed through the chaining of phonological units, and in the visual spatial sketchpad through trajectory through space. Experiment A was designed to require precise tagging of when/where as supported by the episodic buffer and favored deaf signers. Experiment B allowed both when/where tagging and phonological codes. Such dual coding raised overall performance and led to equal performance across groups. Experiment C disabled when/where coding and favored hearing speakers.

Figure 2

In Experiment A participants had to perform a spatial STM task and decide whether the two squares in the probe display were presented in the same location as squares in the sequence (Yes)or whether one of them was in a different location (No). The use of squares diminishes the contribution of the ordered phonological representations and insures greater reliance on spatio-temporal indexing, predicting higher performance in deaf than hearing.

Figure 3

The sequence of stimuli and types of probe displays used in Experiment B were identical to those in Experiment A, except that each square was now labeled with a separate letter. As a result, spatio-temporal indexing and ordered phonological representations could both be efficiently used to encode and maintain information in STM in Experiment B.

Figure 4

The sequences of stimuli used in Experiment C were identical to those in Experiment B. The only difference was in the probe displays in which the two probes were always displayed centrally reducing the contribution of spatio-temporal indexing to the STM task.

Figure 5

Group differences in each experiment demonstrate different biases in deaf signers and hearing speakers. There was a deaf advantage in Experiment A, when only spatial-temporal coding could be used [*=p<.05, one-tailed]. When both spatial-temporal and phonological coding could be used in Experiment B, deaf and hearing spans did not differ. When spatial-temporal coding at recall was disabled in Experiment C, there was a hearing advantage [**= p<.05, two-tailed].