Different types of working memory binding in epilepsy patients with unilateral anterior temporal lobectomy
Introduction
It is well established that the medial temporal lobe, specifically the hippocampus, is essential for long-term memory formation and retrieval (Squire, 2009). However, it is less clear if and how the medial temporal lobe is involved in working memory (WM) processes. It is generally assumed that WM and long-term memory are distinct memory systems, since double dissociations have been described in patients with severely impaired WM, but intact long-term memory and vice versa (Scoville and Milner, 1957, Shallice and Warrington, 1970, Warrington and Weiskrantz, 1970). These memory systems have their own neural correlates, with long-term memory being represented in the medial temporal lobe and WM being represented in a fronto-parietal network (Curtis, 2006, Shallice and Warrington, 1970, Squire, 2009, Warrington and Weiskrantz, 1970).
More recently, results from both patient and neuroimaging studies have led to some discussion concerning this strict distinction of memory systems (for a review, see Ranganath & Blumenfeld, 2005). For example, a lesion study by Olson, Page, Moore, Chatterjee, and Verfaellie (2006) presented evidence for medial temporal lobe involvement in a WM task. Participants had to actively maintain three objects, locations or object–location associations over short delays (i.e., 1 s or 8 s). Patients with bilateral lesions of the medial temporal lobe performed significantly worse than controls in maintaining object–location associations. In addition, neuroimaging studies documented hippocampal activity when participants were actively maintaining object–location associations (Luck et al., 2010, Mitchell et al., 2000a, Piekema et al., 2006).
So far, most WM binding studies focused on object–location binding. Although Olson et al. (2006) did not find a location-only deficit, it could be argued that the spatial character of the task, rather than associative processes per se, contribute to medial temporal lobe involvement. There is increasing evidence that medial temporal lobe involvement during WM tasks is not only material-specific (e.g. spatial vs non-spatial) but also depends on the type of binding (Parra et al., 2011, Piekema et al., 2010). Therefore, we included non-spatial conditions. We designed a WM task based on the studies by Olson et al. (2006) and Mitchell et al. (2000a) to examine whether the medial temporal lobe is involved in WM binding in general, or in specific types of binding only.
There are several ways to distinguish different types of binding. Here, we use the same taxonomy as used in Parra et al. (2009) who distinguish conjunctive binding and relational binding. Conjunctive binding refers to the association between objects and their features, which results in a blended representation of features. Associating an object and its colour is an example of conjunctive binding. Another term for conjunctive binding would be unitized binding (Cohen et al., 1997, Eichenbaum et al., 1994). Relational binding refers to the association between different individual features or items. For example, we refer to relational binding when associating an object with another object or with its location.
To our knowledge, only one patient study systematically compared different types of WM binding (Braun et al., 2011). They found that patients with right hippocampal lesions performed normally on two types of conjunctive binding (colour–shape and colour–letter binding), but were impaired in spatial relational binding (in this case colour–location binding). This may suggest that the medial temporal lobe is involved only in relational binding and not in conjunctive binding. However, an alternative explanation may be that spatial components drive this effect, which may even be attenuated as only patients with lesions in the right hippocampus were included (Milner, Johnsrude, & Crane, 1997).
Our task compares four conditions: conjunctive binding (object–colour binding), spatial relational binding (object–location binding), non-spatial relational binding (object–object binding), and single items. The task was performed by a group of patients who had undergone a neurosurgical treatment for their intractable temporal lobe seizures. The major advantage of these patients is that they all have selective lesions to the anterior temporal lobe. That means that any deficits observed can be attributed to this neural region with more certainty than when patients with other etiologies (like Alzheimer’s disease, where more global lesions are found) are concerned. However, it could also be argued that a history of long-lasting epilepsy can result in a functional reorganisation of the neural substrates underlying memory functions (Braun et al., 2008, Spencer, 2002). We included both patients with left and patients with right-sided lesions, making it possible to evaluate possible lateralization effects.
With respect to relational binding, we expect patients to show impaired performance on spatial relational binding, in line with the studies described above (Braun et al., 2011, Mitchell et al., 2000a, Olson et al., 2006). Non-spatial relational binding was assessed using house–face associations, as an fMRI study demonstrated that house–face associations activated the medial temporal lobe (Piekema et al., 2009, Piekema et al., 2010). Based on these fMRI results, we expect that patients will also show impaired performance on this type of binding.
Concerning the role of the medial temporal lobe in conjunctive binding, mixed results have been found. Previous studies found no evidence for hippocampal involvement during object–colour binding (Braun et al., 2011, Piekema et al., 2006, Piekema et al., 2010). Associating an object with its colour may be a low-level perceptual process achieved earlier in the visual stream, demanding little resources (Piekema et al., 2006, Piekema et al., 2010, Rossi-Arnaud et al., 2006). In addition, Cohen and colleagues showed that blended or unitized representations of features, as opposed to non-unitized relational bindings, can be created by structures outside the hippocampus (Cohen et al., 1997, Eichenbaum et al., 1994). Conversely, one study demonstrated deficits in object–colour binding in patients with Alzheimer’s disease, who also have profound hippocampal dysfunction (Parra et al., 2009).
Finally, in addition to the three binding conditions, a single-item condition was included. Based on previous studies, patient performance was expected to be unimpaired in this condition (Braun et al., 2011, Olson et al., 2006, Parra et al., 2009, Piekema et al., 2006). In contrast, some studies have found deficits in WM for single items. For example, studies in amnesic patients have demonstrated WM deficits for single items for delay periods longer than 6 s (Buffalo et al., 1998, Nichols et al., 2006). A long delay length may cause an overload of WM capacity, which then results in long-term encoding processes that may activate the medial temporal lobe (Jeneson and Squire, 2012, Jeneson et al., 2012). For these reasons we varied delay periods in all conditions (3 or 6 s).
In summary, this study investigates WM for single items and three types of binding in patients who underwent an anterior temporal lobectomy. We examine the hypothesis that only relational binding involves the medial temporal lobe. This means that patients are expected to show unimpaired performance on the single-item condition as well as the conjunctive (object–colour) binding condition. In contrast, patients are expected to show deficits in both types of relational binding, that is spatial (object–location binding) and non-spatial (object–object binding) relational binding. In line with previous studies, we may also expect an effect of delay, in that patient performance would be worse for items and conjunctions maintained for longer delay periods (i.e. 6 s).
Section snippets
Participants
A total of 43 patients and 20 controls were tested. Data from one patient and one control were excluded from further analyses, because they performed more than two standard deviations below the group average. All data from the remaining 42 patients and 19 controls were included in the analyses. All patients (37 right-handed and 5 left-handed) were recruited from Epilepsy Centre Kempenhaeghe, the Netherlands and had undergone a neurosurgical treatment for their intractable temporal lobe
Results
The groups did not significantly differ with respect to age, education level and intelligence level (all p > .10). The patient groups (MTL-L vs. MTL-R) did not significantly differ with respect to absence of postsurgical seizures, epilepsy duration and use of anti-epileptic drugs (all p > .10). Small, yet significant differences were found for sex distribution (p = .02) and age at seizure onset (F(1, 39) = 4.22, p = .047). Post-hoc analyses revealed that the MTL-L group had more male than female
Discussion
The present study investigated different types of WM binding in patients with unilateral anterior temporal lobectomy as treatment of intractable temporal lobe seizures. Importantly, our patients were not densely amnesic (see also the scores within the normal range on the Delayed Memory Index from the WMS-R (Wechsler, 1987), Table 1), which is why any effects that we find cannot be attributed to an overall memory problem. We expected patients to show deficits in relational binding, but not
Acknowledgments
Funding: this study was funded by a VIDI innovational Grant from the Netherlands Organisation for Scientific Research (NWO, No. 452-008-005), awarded to the last author. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
References (50)
- et al.
Distinct neural correlates of associative working memory and long-term memory encoding in the medial temporal lobe
Neuroimage
(2012) Prefrontal and parietal contributions to spatial working memory
Neuroscience
(2006)- et al.
Prose recall and amnesia: More implications for the episodic buffer
Neuropsychologia
(2005) - et al.
The right parahippocampal gyrus contributes to the formation and maintenance of bound information in working memory
Brain and Cognition
(2010) - et al.
FMRI evidence of age-related hippocampal dysfunction in feature binding in working memory
Cognitive Brain Research
(2000) - et al.
Specific deficit of colour–colour short-term memory binding in sporadic and familial Alzheimer’s disease
Neuropsychologia
(2011) - et al.
The right hippocampus participates in short-term memory maintenance of object–location associations
Neuroimage
(2006) - et al.
Doubts about double dissociations between short- and long-term memory
Trends in Cognitive Sciences
(2005) - et al.
The multi-component model of working memory: Explorations in experimental cognitive psychology
Neuroscience
(2006) - et al.
Symmetry and binding in visuo-spatial working memory
Neuroscience
(2006)
When should temporal-lobe epilepsy be treated surgically?
Lancet Neurology
Assessment of working memory in patients with mesial temporal lobe epilepsy associated with unilateral hippocampal sclerosis
Epilepsy & Behavior
The interaction of working memory performance and episodic memory formation in patients with Korsakoff’s amnesia
Brain Research
Feature binding and attention in working memory: A resolution of previous contradictory findings
The Quarterly Journal of Experimental Psychology
Working memory: Theories, models, and controversies
Annual Review of Psychology
Anatomy of the episodic buffer: A voxel-based morphometry study in patients with dementia
Behavioural Neurology
Reorganization of associative memory in humans with long-standing hippocampal damage
Brain
Lesions affecting the right hippocampal formation differentially impair short-term memory of spatial and nonspatial associations
Hippocampus
The human perirhinal cortex and recognition memory
Hippocampus
Memory for items and memory for relations in the procedural/declarative memory framework
Memory
The hippocampus: A “working memory” structure? The effect of hippocampal sclerosis on working memory
Memory
Two functional components of the hippocampal memory system
Behavioral and Brain Sciences
The medial temporal lobe and recognition memory
Annual Review of Neuroscience
The role of the hippocampus in retaining relational information across short delays: The importance of memory load
Learning & Memory
Intact working memory for relational information after medial temporal lobe damage
Journal of Neuroscience
Cited by (0)
- 1
These authors contributed equally to this work.