Distraction-spanning sustained activity during delayed recognition of locations
Introduction
It is well established that spatial working memory tasks recruit activity in a widely distributed network of cortical and subcortical regions (e.g., Corbetta et al., 2002, Jonides et al., 1993, LaBar et al., 1999). What is less clear, however, is which of the many regions identified in functional neuroimaging studies might make necessary contributions to this behavior. Particularly controversial have been the roles of various regions of frontal cortex. Some early neuroimaging studies implicated human dorsolateral prefrontal cortex (PFC) as an important site for the domain-specific retention of location information (Belger et al., 1998, Courtney et al., 1996, McCarthy et al., 1996), results that echoed an influential model of the organization of working memory function in the monkey PFC (Goldman-Rakic, 1987, Wilson et al., 1993). These were followed by several studies that failed to find evidence for domain segregation of lateral PFC working memory activity (e.g., D'Esposito et al., 1998, Nystrom et al., 2000, Owen et al., 1998, Postle et al., 2000b). The implications of this debate were, and continue to be, broader than the narrow brain-mapping question of where different working memory functions are performed, because underlying it are two very different conceptions of working memory. The memory systems view holds that domain-segregated PFC working memory-related activity corresponds to the storage buffers of the multiple-component model of working memory (Baddeley and Hitch, 1974, Baddeley and Logie, 1999). By this account, working memory is supported by specialized systems of the mind and brain, just as visual perception is supported by a visual system (e.g., Courtney, 2004). The emergent processes view, in contrast, sees working memory as a function that arises from the activation, via attention, of systems that have evolved to accomplish perceptual-, representational-, and action-related functions (Postle, in press(b)). By this latter account, spatial working memory can be produced by spatial selective attention (Awh et al., 1998, Awh et al., 2000) and/or by motor preparation (Postle and D'Esposito, 2003, Postle et al., in press, Theeuwes et al., 2005). Further, it holds that delay-period activity of the PFC typically does not reflect the operation of storage processes, but rather, the operation of general purpose control processes (see, e.g., Johnson and Hirst, 1993, Lebedev et al., 2004, Postle, in press(a), Rose and Colombo, 2005).
The neuroimaging studies reviewed up to this point were ill-suited to resolve the debate over the functional organization of visual working memory because they relied on blocked designs that do not permit isolation of specific cognitive components of interest (Friston et al., 1996, Postle and D'Esposito, 2000, Zarahn et al., 1997). They were followed by a second generation of functional magnetic resonance imaging (fMRI) studies employing event-related designs that are capable, in principle, of isolating delay-period activity, a signal that is a candidate neural correlate of storage in short-term and working memory. Early among reports of these event-related fMRI studies was one that argued that the frontal area “specialized” for spatial working memory storage is not in dorsolateral PFC (dlPFC) but, instead, is in the portion of the Superior Frontal Sulcus (SFS) immediately rostral to the Frontal Eye Fields (FEF, Courtney et al., 1998). A study from a different group produced evidence that, consistent with some earlier studies, implicated the dlPFC (Leung et al., 2002). Several subsequent studies have produced data both consistent with (e.g., Leung et al., 2004, Munk et al., 2002, Rama et al., 2004, Sala et al., 2003, Slotnick, 2005) and inconsistent with (e.g., Passingham and Rowe, 2002, Postle, 2005, Postle and D'Esposito, 1999, Postle et al., 2000a) these updated memory systems accounts of spatial working memory.
And so the literature on the cognitive and neural bases of spatial working memory is inconclusive. One reason for this is that understanding of the nature of the information that is being represented by delay-period activity in any given task is a complex undertaking. Possibilities arising from eletrophysiological studies of nonhuman primates include spatial information itself (the interpretation compatible with memory systems views, e.g.,Constantinides and Procyk, 2004, Constantinides et al., 2001, Funahashi et al., 1993), motor preparation (Fukushima et al., 2004, Takeda and Funahashi, 2002, Takeda and Funahashi, 2004), and covert spatial attention (Lebedev et al., 2004). These possibilities have each also been invoked in interpretation of delay-period activity in the human (e.g., Curtis and D'Esposito, 2003, Curtis et al., 2004, Leung et al., 2004, Passingham and Sakai, 2004). A second reason is that, despite the improvement that event-related designs represented over the earlier blocked-design studies, the inferential scope of event-related fMRI studies, too, is limited by the inherently correlational nature of cognitive neuroimaging.
The study reported here employed an experimental procedure intended to permit stronger inference with fMRI data than has been afforded by the neuroimaging studies reviewed up to this point. It applied the logic that sustained activity that is necessary for spatial working memory must persist across intervening distractors, whereas activity that is not necessary may be “filtered out” by these distractors. More specifically, it used an ABCA design, in which a trial could require evaluation of one (an “AB” trial), two (an “ABA” trial), or three memory probes against the target stimulus. Retention of activity across the three delay periods of this task would be a necessary (although not sufficient) condition that a region must meet if it were to be considered necessary for successful working memory performance. The present study can be seen as a companion to a previous study requiring working memory for faces. This previous study found that only posterior fusiform cortex supported delay-period activity that was reliably sustained across three delay periods (Postle et al., 2003).
In the present study, we predicted that we would find distractor-spanning sustained delay-period activity in several cortical regions, including the FEF, Intraparietal Sulcus (IPS), and Superior Parietal Lobule (SPL), and perhaps also in caudate nucleus. Importantly, we also predicted that we would not find evidence for distractor-spanning delay-period activity in dlPFC or in posterior SFS. We made these predictions for two reasons. One derives from the fact that spatial working memory and spatial selective attention share largely overlapping networks (e.g., Corbetta et al., 2002, LaBar et al., 1999), and the spatial attentional networks that overlap with working memory are generally not found anterior to the FEF (e.g., Corbetta et al., 1998, Corbetta et al., 2002, Kim et al., 1999, Yantis and Serences, 2003). Additionally, evidence for the “attention-based rehearsal” of spatial information has been found in posterior, but not frontal, cortical regions (Awh et al., 2000, Postle et al., 2004). The second reason for our predictions is that our previous direct tests have been unable to dissociate spatial delay-period activity from object delay-period activity (Postle and D'Esposito, 1999) or from oculomotor control-related activity (Postle et al., 2000a) (see also Postle, 2005, Slotnick, 2005), leading us to hypothesize that the delay-period activity of neither dlPFC nor posterior SFS is necessary for spatial working memory.
Section snippets
Subjects
Our methods were approved by the Health Sciences Institutional Review Board of the University of Wisconsin–Madison. Sixteen healthy young adults who reported no history of neurological or psychiatric disorders, and no recent use of psychoactive drugs, participated after giving informed consent. The fMRI data from 3 subjects were discarded due to excessive movement in the scanner.
Materials and apparatus
Stimuli were presented, and responses collected, on a PC running Eprime software. Stimuli were white circles of
Behavioral
Accuracy declined as a function of the number of delay periods (F(2,30) = 15.9; P < 0.0001). Reaction time (RT), on the other hand, did not vary with trial type (F(2,30) = 1.5; n.s.) (Table 1). To validate the assumption that subjects used spatial information to perform the task (as opposed to, for example, an internally generated verbal code), we evaluated performance on nonmatching trials as a function of the distance between the target stimulus and the trial-final probe stimulus. This
Discussion
Consistent with our predictions, and with many previous studies, we found location delay-related activity in a broadly distributed network of cortical regions, including frontal and parietal regions associated with attention and oculomotor control. Across three types of analyses, these regions were seen to support sustained, distraction-spanning signals that may correspond to the mnemonic retention of the location of the target stimulus. Also consistent with our predictions was the absence of
Acknowledgments
The author thanks Olufunsho Faseyitan and Craig Rypstat for the programming support, and O.F., Christopher Jordan, and Andrew Nick for the assistance with data collection, processing, and analysis. Supported by NIH MH064498.
References (79)
- et al.
The variability of human, BOLD hemodynamic responses
NeuroImage
(1998) - et al.
Working memory
- et al.
A common network of functional areas for attention and eye movements
Neuron
(1998) - et al.
Persistent activity in the prefrontal cortex during working memory
Trends Cogn. Sci.
(2003) - et al.
Functional MRI studies of spatial and non-spatial working memory
Cogn. Brain Res.
(1998) - et al.
Analysis of fMRI time-series revisited
NeuroImage
(1995) - et al.
The trouble with cognitive subtraction
NeuroImage
(1996) - et al.
Variation of BOLD hemodynamic responses across subjects and brain regions and their effects on statistical analyses
NeuroImage
(2004) - et al.
The large-scale neural network for spatial attention displays multifunctional overlap by differential asymmetry
NeuroImage
(1999) - et al.
Neuroanatomic overlap of working memory and spatial attention networks: a functional MRI comparison within subjects
NeuroImage
(1999)