Representing spatial and ordinal information in working memory
Mehdi Senoussi (1) and Wim Gevers (2)
(1) UGent; (2) ULB
Many empirical and modelling efforts have been invested to study how information is represented in working memory (WM) at computational, behavioral and neural levels. There is consensus that the maintenance of spatial, item and ordinal information relies on both independent and shared mechanisms and neural substrates. In this symposium we gather different views and accounts on how different types of information are represented and maintained in working memory. Furthermore, we want to promote a more integrated picture of how different facets of WM are supported by common mechanisms. To encourage this approach, we invited speakers from different institutions in Belgium and Germany, and investigate different sub-fields of WM such as spatial WM and attention, ordinal WM, and developmental and clinical aspects of WM.
Speaker 1: Precision of ordinal representation in working memory: domain specific or domain general?
Lucie Attout (1), Nathan Leroy (1) and Steve Majerus (1)
Ordinal processing involves processing of the sequential relations between stimuli of a stimulus set. This crucial ability has been studied extensively and separately in different domains such as working memory (WM) and numerical cognition. Several behavioural and neuroimaging studies suggest the possibility of common ordinal coding mechanisms across different domains. This fMRI study assessed the hypothesis of common ordinal representational mechanisms across the WM, the number and the letter domains. We administered three ordinal judgement tasks (for alphabetical, numerical, and verbal WM judgment) with further manipulation of ordinal distance, and a luminance judgment control task. Results showed above-chance-level classification of ordinal distance not only within but, critically, also between tasks. Luminance distance could not be decoded by an ordinal distance classifier, indicating that the results are specific to ordinal processing, and not to distance per se. These results provide support for theoretical accounts considering the existence of domain general serial order coding mechanisms, at least for explicit ordinal judgment.
Speaker 2: The interplay between spatial processes and verbal working memory
Muhammet Ikbal Sahan (1), Vesal Rasoulzadeh (1), Giulia Cristoforetti (1), Jean-Philippe van Dijck (1,2) and Wim Fias (1)
(1) UGent; (2) Thomas More
Classic accounts on working memory (WM) distinguish verbal from visuospatial WM. Yet, recent work has shown that spatial processes are also involved in verbal WM in that, begin elements are associated with the left side of space and end elements with the right side. Our aim was to gain direct access to the spatial search processes involved in verbal WM using state-of-the-art eye-tracking. Eye-movements were continuously tracked while participants performed an auditory go/no-go task during which free-viewing was allowed. Participants were instructed to detect a beep that was either presented to the left or right ear. Each beep was preceded by a binaurally presented digit serving as a go/no-go cue: the beep was to be detected only when this cue was part of a serially encoded WM set (i.e., 4-digit sequence). We hypothesized that cues from the beginning of a sequence would trigger gaze shifts to the left side of space and cues from the end to the right. Indeed, the eyes were spontaneously moving as a function of the ordinal position in WM: a leftward shift was observed when participants heard a cue from the beginning of the sequence and a relatively more rightward shift was observed when a cue was heard from the end of the sequence. Our findings suggest that order information is spatially grounded in WM and that search through memory resembles visual search. With the eyes as a window to the mind, our study provides a promising novel approach to understand verbal thought processes.
Speaker 3: Visual attention in complex environments: breadth of attention and its link to other cognitive functions
Daniel Memmert (1)
(1) German Sport University Cologne
Abstract to follow
Speaker 4: Comparing and priming ordinal information in Working Memory
Dewulf Myrtille (1), Antoine Sophie (1) and Gevers Wim (1)
Letters are ordinally coded in long-term memory. We easily understand that F precedes H but follows B. A comparison distance effect (CDE) is observed when letters are compared to a fixed reference letter (e.g. before or after O? Gevers et al, .2006; Van Opstal et al., 2008): reaction times decrease with increasing alphabetical distance. However, no priming distance effect (PDE) is observed with letters of the alphabet. Van Opstal and colleagues presented letter primes just before the onset of a target letter that had to be compared to a fixed standard. Letter primes did not influence performance on the targets. It can be concluded that ordinal positions in LTM do not representationally overlap. But what about ordinal positions in working memory? Research suggests that ordinal positions in working memory are spatially coded (van Dijck et al., 2011). The possibility thus exists that more representational overlap exists for close compared to far ordinal positions. To test this, a similar task as in Van Opstal et al was designed using letters and both priming and comparison distance were manipulated. The crucial difference was that the ordinal positions of the letters were encoded in working memory. A CDE was observed as expected. Crucially, also a PDE emerged. Finally, the PDE and the CDE were unrelated to one another. This latter finding strengthens the earlier made suggestion that CDE and the PDE reflect different underlying processing mechanisms. Most importantly, we demonstrate representational overlap of ordinal positions in working memory while this is not the case in LTM.