پردازش اطلاعات مکانی-زمانی در اسکیزوفرنی شروع اولیه: بررسی میزان حافظه کوتاه مدت و استعداد آن برای حواس پرتی

Abstract Memory impairment is a core feature in schizophrenia (SZ). The aim of this study was to investigate short-term memory (STM) and its sensitivity to distraction with visual–spatial material. This study comprised 23 recent-onset SZ patients and 23 healthy controls. The degree of disruption upon recall from interleaving irrelevant items within a sequence of to-be-remembered items—the sandwich effect [Hitch, G. J. (1975). The role of attention in visual and auditory suffix effects. Memory and Cognition, 3, 501–505]—was examined. STM performance, whether in the presence or absence of distraction, was poorer and markedly more vulnerable to disruption in SZ. Our results suggest that processing spatial information in STM is susceptible to interference in SZ.

Introduction Many researchers view attentional control and short-term memory (STM) processing dysfunctions in schizophrenia (SZ) as core features of the disorder. However, most, if not all, of the work on STM and its interaction with selective attention has focused on processing of verbal items. Yet, processing of spatial information seems to be particularly affected in SZ (Park & Holzman, 1992) and it is not clear whether dysfunctional attentional processes are responsible, at least in part, for the observed deficits in spatial STM (see Lee & Park, 2005, for a discussion). In the current study, we examine the capacity to remember temporal–spatial information over the short-term and its susceptibility to distraction within a paradigm well established in experimental cognitive psychology: the sandwich paradigm (Hitch, 1975 and Tremblay et al., 2005). The interplay of STM and selective attention is usually investigated through the manipulation of distraction, either present or absent when a sequence of to-be-remembered (TBR) items is learned. Oltmanns and Neale (1975) developed a procedure known as the Digit Span Distractibility Test (DSDT), in which irrelevant items (digits said in a female voice) are interleaved with TBR items (digits read in a male voice). If, for serial recall of spoken lists, irrelevant items are interpolated between TBR items, there is a significant increase in recall errors for SZ patients as well as for healthy participants (e.g., Corrigan and Green, 1991, Harvey and Pedley, 1989, Oltmanns, 1978, Oltmanns and Neale, 1975 and Rund, 1989). However, this limitation is more pronounced in SZ. Typically, SZ patients recall fewer items than controls whether the TBR list is presented with or without interleaved distractors but the presence of distractors has a much greater impact on recall for SZ patients. Indeed, for a list of six digits, the magnitude of disruption can reach up to 46% in the recall performance of SZ patients and up to 26% for healthy subjects ( Frame & Oltmanns, 1982; see Spring, Weinstein, Freeman, & Thompson, 1991, for a meta-analysis). In the study of normal cognition, a procedure similar to that of the DSDT is referred to the sandwich effect paradigm (e.g., Hitch, 1975). The mere interpolation of auditory irrelevant items within a sequence of TBR items produces a robust but small deleterious effect upon serial recall (e.g., Baddeley et al., 1993 and Nicholls and Jones, 2002). The so-called sandwich effect has recently been replicated with lists of visual–spatial items (Tremblay et al., 2005), using the dot task, regarded as a good analogue of the typical verbal serial recall task and a procedure that minimizes the possibility that spatial information be verbally recoded (see Jones, Farrand, Stuart, & Morris, 1995). When seeking if a verbal memory phenomenon can be reproduced in the spatial domain it is critical to ensure that the TBR information is truly spatial and non-verbal. In SZ, there is a relative dearth of studies on spatial STM, especially for temporal–spatial information and the effect of visual–spatial distraction. From reviewing the literature, it seems consistent across studies that spatial STM is impaired (but see Salame, Danion, Peretti, & Cuervo, 1998, for a different view), though the variability in the procedures is considerable. There is ample evidence that processing of spatial information is dysfunctional as observed in variants of delayed response and delayed perceptual discrimination tasks (e.g., Park and Holzman, 1993 and Tek et al., 2002). Typically, in such tasks, a target is briefly displayed, followed by a delay, and then, participants have to respond by locating or detecting the target. Those tasks are very different to serial recall procedures in terms of memory load and response requirements. With regard to serial recall or sequence reproduction of spatial material (e.g., the Corsi block task), most researchers have observed marked deficits (Fleming et al., 1997, Fraser et al., 2004 and Stone et al., 1998), while some others failed to find a significant difference between SZ patients and controls (see Clare, McKenna, Mortimer, & Baddeley, 1993). However, the balance of evidence weighs strongly for a genuine impairment, and thus supports the claim that the deficit in STM processing is not restricted to verbal material (see, e.g., Chey, Lee, Kim, Kwon, & Shin, 2002). According to Fischer (2001) there are two key components in spatial serial recall procedures that contribute to the memory load: serial order and spatial location. This observation also applies to verbal serial recall, which requires processing item and order information. In the case of spatial material, the location of the blocks (Corsi block task) or dots (dot task) represents item information. Using a computerized version of the Corsi block task, Dreher and his collaborators (2001) developed variants of the task that selectively tap either on spatial recall (item) or temporal recall (order). They obtained a significant difference in performance between the SZ group and the healthy subjects on the spatial but not on the temporal version of the task. The latter pattern of results observed in the spatial domain suggests that item memory is altered in SZ rather than order memory (see also Fraser et al., 2004); which parallels with the finding that order memory is preserved in verbal serial recall (Elvevag, Weinberger, & Goldberg, 2001). To our knowledge, the impact of visual–spatial distraction on STM for visual–spatial information has never been tested in SZ. In the case of verbal serial recall, there is ample evidence that SZ patients are more sensitive to distraction than controls. One key question is whether this pattern of distraction with recall also extends to spatial STM. Based on what is probably the most influential model of STM, the so-called Working Memory (WM) model (Baddeley, 2003; see also Baddeley & Hitch, 1974), one could expect to find a different pattern of distraction for spatial STM. A key assumption of the WM model is that STM is divided into separate subsystems such as the visual–spatial sketchpad and the phonological loop (respectively responsible for processing visual images and verbal material). This view is based around the premise that the cognitive architecture is divided into separate modules, such as verbal or spatial components. However, there is accumulating evidence, that, at least for processing serial order, spatial and verbal STM are functionally equivalent. Both show similar serial position curves, with primacy and marked recency effects (e.g., Jones et al., 1995 and Smyth and Scholey, 1996) and are sensitive to interference effects or distraction (e.g., Tremblay et al., 2001 and Tremblay et al., 2005). In order to test whether the susceptibility of STM to distraction, for spatial material, is more pronounced in SZ than in healthy controls, we propose to employ the visual–spatial version of the sandwich paradigm. This will allow us to determine whether the type of deficits reported for verbal STM (e.g., Corrigan and Green, 1991, Harvey and Pedley, 1989, Oltmanns, 1978, Oltmanns and Neale, 1975 and Rund, 1989) is also found with spatial material, that is, a greater impact of distraction in patients with SZ in comparison with healthy controls. In addition, we chose to test the previously mentioned hypotheses with recent-onset participants to minimize the risk of a prevalence bias (i.e. the tendency for prevalent cases being more likely to include an overproportion of more severe cases more likely to show more severe neuropsychological impairment; Lehoux et al., 2003 and Roy et al., 2003).

3. Results Participants’ responses were scored according to a strict serial recall criterion. For each group, the mean number of correct responses at each serial position as a function of condition is depicted in Fig. 2. A visual inspection of the reported means suggests that the level of recall of SZ patients was poorer than that of healthy controls and the impact of distraction is more pronounced in SZ. A mixed ANOVA was conducted on the data with respect to condition (2 levels: control and sandwich) and serial position (5 levels) as within-subjects variables, and group (2 levels: SZ patients and healthy controls) as a between-subjects variable. A Greenhouse–Geisser correction was used for adjusted values when the assumption of sphericity was not met. The main effect of group was significant, F(1, 44) = 22.53, MSE = 35.43, p < .001, which means that the level of recall performance of SZ patients was poorer than that of healthy controls. The main effect of condition was significant, F(1, 44) = 11.15, MSE = 3.93, p < .01, revealing that visual–spatial STM is susceptible to visual interference. The main effect of serial position was also significant, F(4, 176) = 18.67, MSE = 2.44, p < .001, which suggests the presence of a serial position curve, with primacy and recency effects. The interaction between condition and serial position, F(4, 176) = 2.20, MSE = 2.94, p = .12, was not significant, which mean that all serial position seem vulnerable to distraction—as is the case with the classical sandwich effect ( Tremblay et al., 2005). The interaction between serial position and group, F(4, 176) = 5.77, MSE = 2.44, p < .01, was significant. The latter interaction may reflect a performance near ceiling for the group of healthy controls. No further theoretical significance is attached to these interactions. The triple interaction was not significant, F(4,176) = .55, MSE = 2.94, p = .57. Dot task: mean number correct as a function of serial position for control and ... Fig. 2. Dot task: mean number correct as a function of serial position for control and distraction conditions for healthy controls (a) and SZ patients (b). Maximum number of corrects is 15. Error bars represent standard errors. Figure options With regards to the impact of distraction on STM performance for healthy controls compared to that of the SZ patients, the key interaction of condition and group was significant, F(1, 44) = 4.48, MSE = 3.93, p < .05. Given our a priori prediction of an increased disruptive effect of distraction for SZ patients, we decomposed the interaction by carrying out simple main effect tests on the data (alpha of .0125 as corrected with the Bonferroni procedure). Disruption caused by the presence of irrelevant stimuli was of greater magnitude in SZ patients, t(22) = 3.04, p < .01, Cohen’s d = 0.39, compared to the group of healthy controls, t(22) = 1.38, p = .18, Cohen’s d = 0.22. The simple main effect tests also revealed a significant difference between the two groups for the control condition, t(44) = −3.96, p < .001, as well as for the sandwich condition, t(44) = −5.02, p < .001. Memory performance, with and without distraction, was poorer with SZ patients. There was a significant difference between SZ patients and healthy controls on the estimated IQ, t(44) = −3.60, p < .01 (see Table 1). It is likely that abbreviated IQs obtained in the present study for the SZ patients (109.74) are overestimated. There is some evidence to suggest that dyadic short forms of IQ testing may overestimate IQ ( Allen et al., 1997 and Ringe et al., 2002) by an average of 8 points in SZ patients when using the dyad of vocabulary and matrix reasoning ( Ouellet et al., submitted).