Comparing the immediate free recall of verbal and visuo-spatial stimuli:List length, capacity and output order effects in single- and dual-modality tasks

When participants are presented with a list of words, and are asked to recall the items in any order (Immediate Free Recall; IFR), they tend to initiate their recall with the first list item when presented with short lists and with one of the last four items in longer lists. Chapter 2 examined whether this tendency necessitates a language-based retrieval mechanism by replicating and extending this finding in verbal and visuo-spatial IFR. The observed similarities between the two modalities are argued to be reflective of either a domain-general retrieval mechanism that operates on all stimuli at all timescales, or two domain-specific mechanisms that operate in quasi-identical ways. To distinguish between these two possibilities, Chapter 3 compared capacity and output order effects in both stimulus domains in single- and dual-modality IFR tasks. The number of items recalled in dual-modality IFR suggest partially independent capacities, but the output orders across the two modalities were greatly constrained; findings that cannot be fully explained through either a domain-general or domain-specific framework. Additionally, participants' tendency to alternate across modalities may be due to one-to-one associations of auditory words with the contemporaneous visuo-spatial locations. Consequently, Chapter 4 examined whether this alternating output strategy is still present when list structure is completely randomised and the items temporally off-set. I argue that this tendency is not entirely due to some form of binding but may, at least in part, reflect the most efficient way of outputting a mixed-modality list. Moreover, the asymmetry in the recall accuracy of the two modalities is due to increased number of to-be-recalled stimuli rather than increased output interference. I suggest that overall, a domain-general approach, such as the Embedded Processes model (Cowan, 2005), coupled with a forward-ordered retrieval mechanism based on temporal grouping (Farrell, 2012) is better placed to account for these novel findings.