Dissociating habit and recollection: evidence from Parkinson’s disease, amnesia and focal lesion patients
Introduction
In recent years, there has been great interest in dissociating different types of memory and the brain areas that may mediate different memory processes or systems. Converging research has supported the distinction between two qualitatively different types of memory based on evidence from cognitive psychology [21], [33], [59], patient populations [7], [35] and animal learning [34]. Declarative, or explicit, memory has been characterized as a conscious, intentional and effortful ability to recollect a previous episode. This type of memory has consistently been shown to be impaired following damage to medial temporal lobe (MTL) structures [56] and also after frontal structures if more strategic processing is implicated at encoding or retrieval [35]. In contrast, nondeclarative, or implicit, memory has been characterized as an unconscious, automatic basis of responding that does not rely on the ability to recollect. This type of memory is typically preserved in amnesia (for reviews see [37], [57]) and is unaffected by other experimental manipulations that impair conscious remembering (e.g. [65]). As we describe findings from the neuropsychological patient literature, we will generally refer to these two types of memory as declarative and nondeclarative, respectively, although the terms ‘explicit’ and ‘implicit’ could be interchanged with these. In this study, we more specifically examine declarative and nondeclarative memory by investigating recollection and habit. The terms ‘recollection’ and ‘habit’ are more precise terms that are defined within the context of Jacoby’s process-dissociation framework [21], as described later. By using a test of memory in which the contribution of recollection and habit could be determined, we hoped to assess the role of different brain structures in mediating recollection and habit.
There have been numerous conceptualizations of nondeclarative memory and it has been measured using a variety of techniques. Nondeclarative memory has been expressed as procedural learning or skill acquisition, measured on motor tasks such mirror reading, maze learning or rotary pursuit [18]. It has also included more cognitive skills such as artificial grammar learning, category learning and probability-matching [27]. Priming has also been subsumed under the broad label of nondeclarative memory. While these different tasks may initially appear dissimilar, it is believed that the cognitive processes that underlie them are all automatic, unintentional processes that operate independently of a conscious effort to learn or remember. Consequently, it is assumed that the neural mechanisms that give rise to this type of automatic responding are different from the neural mechanisms involved in declarative memory. Nondeclarative memory does not appear to involve the hippocampus and associated MTL structures, or the frontal lobes if the tests are perceptual and data-driven [63], [64]. However, our understanding of the brain structures that may mediate nondeclarative memory is limited. The difficulty of localizing nondeclarative memory in the brain may stem from its context-specific nature, as it has been shown to be task-dependent and sensitive to subtle changes in presentation and form (e.g. [22]). Nondeclarative memory has also been assessed with a wide range of different measures, and therefore, it is unlikely to be a single entity consistent across all tasks.
Different forms of nondeclarative memory may be dependent upon distinct neuroanatomic regions [18], [35], [36], [37]. Brain imaging studies that have investigated nondeclarative memory in normal adults have revealed that it is mediated by different brain regions compared to declarative memory [5]. Functional dissociations have also been demonstrated with scalp recordings of event-related brain potentials (ERPs) for declarative and nondeclarative memory [51], as well as recollection and habit [16]. Using positron emission tomography (PET), researchers have shown that visual perceptual priming is mediated by occipital brain areas [56] while others have demonstrated the involvement of left frontal regions in conceptual priming [6], [9], [45], [61]. There is evidence that motor learning tasks recruit subcortical structures such as the basal ganglia (e.g. [12], [26], [46]) and recently, there has been a demonstration of striatal activation in cognitive habit-learning, using functional magnetic resonance imaging (fMRI) [44].
While it has been established that amnesics, who have sustained damage to the hippocampus and related MTL structures, demonstrate striking deficits in declarative memory in the presence of intact priming (e.g. [7], [53]), the opposite dissociation has also been found. That is, patients with Parkinson’s disease (PD) have shown impaired nondeclarative memory in the presence of intact recollection (e.g. [27], [52]). PD is a neurological disorder that causes a degeneration of neurons in the substantia nigra, resulting in a loss of dopaminergic input to the striatum, which includes the caudate nucleus and putamen in the basal ganglia. Individuals with PD without dementia serve as a useful model of striatal dysfunction in the presence of an otherwise intact neurological system. Studies of skill learning in PD patients have revealed nondeclarative memory deficits on motor tasks [1], [18], [62], as well as nonmotor tasks [29], [32], [49], [52]. However, when cognitive measures that do not rely on a motor output are used, PD patients do not always reveal deficits in cognitive skill learning [13]. On tests of implicit memory, PD patients have demonstrated intact word stem repetition priming abilities [2], [18], [20], [30], and artificial grammar learning and category learning have also been preserved in PD patients [48].
One factor that may account for the inconsistent findings of the effects of PD on nondeclarative memory is the role of conscious strategic processes. That is, on nondeclarative tasks that do not have a motor component, such as category learning, artificial grammar learning or priming, participants are not instructed to use any intentional or conscious strategies to complete the task. Sometimes, however, participants become aware of task relationships and start to use intentional strategies to assist them with the task. If PD patients have relatively preserved recollective abilities, then it seems possible that they could also use conscious strategies to assist them on many nondeclarative tasks. This might especially be likely an easy tasks that do not exhaust cognitive resources. The extent to which conscious strategies have contaminated performance on nondeclarative tasks has not been measured and may account for inconsistencies in the literature. Indeed, Knowlton et al. [27] showed that PD patients were initially impaired on a probabilistic habit-learning task over the first 50 trials but the same patients did not perform any differently from controls on later trials. Knowlton et al. argued that conscious strategies affected performance on this nondeclarative task, but only on later trials. They suggested that the PD patients started to use their intact conscious strategies to assist their performance later in the task, ameliorating earlier differences that were apparent when performance was based primarily on habit-learning. These findings suggest that the striatum is involved in the initial stages of habit-learning but that the hippocampus and related MTL structures, and possibly the frontal lobes, may become involved in later stages of task performance. Further evidence to support this interpretation comes from a study by Knowlton et al. [28] in which they tested amnesics and controls on the same probabilistic habit-learning task described above. They found that amnesics did not differ from controls on the first 50 trials but that over time they performed more poorly and a difference emerged. These researchers interpreted this finding by arguing that control patients used conscious strategies to assist them later in the task but that the amnesics lacked such abilities and, therefore, could not assist their performance to the same extent.
The purpose of the current study was to further investigate the role of the striatum and the MTL in memory by examining habit-learning and recollection in patients with PD and amnesia, respectively. Rather than rely on different types of tasks to measure declarative and nondeclarative memory processes, we made use of Hay and Jacoby’s [14] extension of Jacoby’s [21] process-dissociation procedure to investigate the contribution of habit and recollection to performance within a single task. The advantage of our procedure is that it allows us to avoid the contamination concerns associated with previous methods by measuring habit and recollection within a single probability-learning paradigm. We were interested in determining whether PD patients would reveal a deficit in cognitive habit-learning when measured without the influence of conscious strategies. Another issue we wanted to address with our procedure was the extent to which declarative memory is affected by PD. PD patients have demonstrated intact declarative memory (e.g. [3], [27], [52]) however, other researchers have found PD patients to be impaired on declarative memory tasks, especially when they involve more strategic, effortful processing [60]. As PD also affects fronto–striatal connections in the brain, we more closely examined the frontal contribution to habit and recollection in Experiment 2, by examining memory performance in patients with focal frontal lesions, as well as in a patient with focal striatal lesions with no frontal involvement.
The results of Knowlton et al.’s studies [27], [28] suggested that PD patients failed to establish habits but were able to use conscious strategies to assist them in later stages of the task. In contrast, amnesics formed habits initially, but had difficulty using conscious strategies to assist their performance as the task progressed. Based on the Knowlton et al. findings, the following predictions were made in the current study. We predicted that PD patients would be impaired at habit-learning but demonstrate intact recollective abilities. In contrast, we expected the amnesics to demonstrate intact habit estimates but have deficits in recollection.
The first phase of the experiment was a training session designed to create habits of specific strengths. This part of the paradigm was very similar to a traditional two-choice probability-learning experiment. Participants were exposed to pairs of semantically-related words with the probabilities of the pairings varied. A stimulus word was presented with two related responses such that a “typical” response (e.g. knee-bend) occurred twice as often (67%) as an “atypical” response (e.g. knee-bone, 33%). Once a habit was established, the second phase of the experiment was a series of short memory tests. Participants studied short lists of word pairs they had seen earlier in training and then were tested with the stimulus word and a fragment of the target response that could be completed with either response from training (e.g. knee-b_n_). Estimates of recollection and habit were derived by applying the process-dissociation equations to performance in these study–test sessions.
On congruent trials, participants studied items that were either made typical in training and, therefore, participants could respond correctly by either recollecting (R) the item from the short study list, or by relying on their habit (H) of giving the typical response when recollection failed (1−R). Recollection was congruent with the typical habit formed in training. The probability of responding correctly with a typical item on congruent trials can be written as
On incongruent trials participants studied atypical items and, therefore, habit was now a source of error. Incorrectly responding with a typical item occurred if participants failed to recollect the response they had just studied in the preceding list (1−R) and instead relied on their habit of giving the typical response (H). The probability of incorrectly responding with a typical item on incongruent trials can be written as
An estimate of recollection can be calculated by subtracting the probability of responding with a typical response on congruent and incongruent trials
Given an estimate of recollection, an estimate of habit for the typical response can be derived by algebra
By using the process-dissociation equations outlined above, we were able to eliminate contamination concerns typically associated with declarative and nondeclarative tests. As such, we could examine the effects of PD and amnesia on pure estimates of habit and recollection.
Section snippets
Participants
Patients: Twenty-four PD patients participated in the study. The diagnosis of PD without dementia was confirmed by a senior neurologist at the Movement Disorders Centre at the Toronto Hospital in Toronto, Ontario. Based on the Hoehn and Yahr rating scale [19], 12 patients (6 men/6 women) were in the early stages of PD and had ratings of 2.5 or less. The remaining 12 patients (8 men/4 women) were in the moderate to severe range of PD and had ratings of at least 3.0. The mild PD group averaged
Method
Participants: Five patients with focal frontal lesions were recruited for the study from the volunteer pool at the Rotman Research Institute. Four of the patients had undergone brain surgery to have frontal meningiomas removed from anterior portions of their brain (one on the left side, three on the right side). One patient had epilepsy and underwent surgery on her right frontal lobe to help control her seizures. All focal lesions were located in dorsolateral regions of the frontal lobes. The
General discussion
Using Hay and Jacoby’s [14] extension of the process-dissociation procedure, we examined the effects of PD and amnesia on habit and recollection within a single task. We found that the moderate PD group was significantly impaired at habit-learning while the mild PD group and amnesics were not. In addition, a patient with a focal striatal lesion also demonstrated impoverished habit-learning performance. Both the moderate PD group and the striatal lesion patient failed to reveal
Concluding comments
PD is a neurological disorder that primarily affects motor functioning. However, it is becoming more apparent that there are also significant cognitive effects that emerge as the disease progresses that should not be overlooked. The effects of PD on memory have been difficult to discern due to problems arising from contaminated task performance. Separating out memory processes within a single task allowed us to examine the effects of PD on different types of memory. Based on our findings, there
Acknowledgements
This research was supported by the Ben and Hilda Katz Post-Doctoral Fellowship at the Rotman Research Institute of Baycrest Centre to Janine Hay and by a Natural Sciences and Engineering Research Council (NSERC) operating grant to Morris Moscovitch. We are grateful to Dr. Anthony Lang for referring PD patients to us. We thank Eileen Halket and Luisa Del Rizzo for screening PD patients at the Toronto Hospital Movement Disorders Clinic and Heidi Roesler for testing control participants. We also
References (65)
Recall and recognition memory in Parkinson’s disease
Cortex
(1993)- et al.
Cognitive function in Parkinson’s disease: from description to theory
Trends in Neuroscience
(1990) - et al.
Functional–anatomic correlates of object priming in humans revealed by rapid presentation event-related fMRI
Neuron
(1998) - et al.
Memory and skill acquisition in Parkinson’s disease and frontal lobe dysfunction
Cortex
(1995) - et al.
Role of the striatum, cerebellum, and frontal lobes in the learning of a visuomotor sequence
Brain and Cognition
(1997) - et al.
Mini-mental state: a practical method for grading the cognitive state of patients for the clinician
Journal of Psychiatric Research
(1975) A process-dissociation framework: separating automatic from intentional uses of memory
Journal of Memory and Language
(1991)- et al.
Procedural learning in Parkinson’s disease: intact and impaired cognitive components
Neuropsychologia
(1999) - et al.
Perceptual priming in Alzheimer’s and Parkinson’s diseases
Neuropsychologia
(1996) - et al.
Spatial and nonspatial working memory at different stages of Parkinson’s disease
Neuropsychologia
(1997)
Probabilistic learning and reversal deficits in patients with Parkinson’s disease or frontal or temporal lobe lesions: possible adverse effects of dopaminergic medication
Neuropsychologia
Memory for temporal order and conditional associative learning in patients with Parkinson’s disease
Neuropsychologia
Parkinson’s disease patient’s behavior in a covered maze learning task
Neuropsychologia
Negative transfer (AB–AC) learning in normal old people and in patients with unilateral temporal-lobe or frontal-lobe lesions: heightened interference on conceptual, implicit tests of memory
Brain and Cognition
Procedural memory and Parkinson’s disease
Dementia
Implicit and explicit memory in Alzheimer’s disease and Parkinson’s disease
Journal of Clinical and Experimental Neuropsychology
Functional anatomical studies of explicit and implicit memory retrieval tasks
Journal of Neuroscience
Preserved learning and retention of pattern-analyzing skills in amnesia: dissociations of knowing how and knowing that
Science
Semantic encoding and retrieval in the left inferior prefrontal cortex: a functional MRI study of task difficulty and process specificity
Journal of Neuroscience
Functional mapping of sequence learning in normal humans
Journal of Cognitive Neuroscience
Procedural memory in Parkinson’s disease: impaired motor but not visuoperceptual learning
Journal of Clinical and Experimental Neuropsychology
Separating habit and recollection: memory slips, process-dissociations and probability-matching
Journal of Experimental Psychology: Learning, Memory and Cognition
Separating habit and recollection in young and older adults: effects of elaborative processing and distinctive context
Psychology and Aging
Effects of aging and context-specificity on habit and recollection
Brain and Cognition
Neuropsychological evidence for multiple implicit memory systems: a comparison of Alzheimer’s, Huntington’s, and Parkinson’s disease patients
The Journal of Neuroscience
Parkinsonism: onset, progression and mortality
Neurology
Comparison of patients with Alzheimer’s and Parkinson’s Disease on different explicit and implicit memory tests
Journal of Behavioral Neurology
On the relationship between autobiographical memory and perceptual learning
Journal of Experimental Psychology: General
MEM: mechanisms of recollection
Journal of Cognitive Neuroscience
Source monitoring
Psychological Bulletin
Cited by (41)
The contribution of different prefrontal cortex regions to recollection and familiarity. A review of fMRI data
2017, Neuroscience and Biobehavioral ReviewsCitation Excerpt :Instead, in Duarte et al. (2005) familiarity impairment in patients with either left or right prefrontal damage was related to items encoded in the contralesional hemifield. Moreover, in a study with patients who had undergone the resection of right frontal lobe tumors (Anderson et al., 2011), a relationship was reported between familiarity performance and lesion size at the level of the right inferior PFC. The contradictory data in the lesion literature are likely due to the fact that the anatomical subdivisions adopted were too vast and heterogenious (e.g., left vs. right differences in Aly et al., 2011; Duarte et al., 2005; Hay et al., 2002; lateral vs. orbital differences in Wheeler and Stuss, 2003; medial vs. lateral vs. orbital differences in MacPherson et al., 2008). Moreover, in Anderson et al. (2011) most patients’ damage was not clearly circumscribed to distinct prefrontal areas.
The effect of focal cortical frontal and posterior lesions on recollection and familiarity in recognition memory
2017, CortexCitation Excerpt :Additionally, the temporal neocortex mediates higher order sensory processing as it is part of a multimodal network that is activated during vivid recall, suggesting that vivid recall is mediated by the same regions engaged at encoding (Chen et al., 2017). To our knowledge only one study examining subjective recollection has included both patients with focal frontal and more posterior damage (in this case MTL lesions) (Hay et al., 2002). While the two groups of patients were not directly compared to each other, both groups were reported to have deficits in recollection.
Evidence of an amnesia-like cued-recall memory impairment in nondementing idiopathic Parkinson's disease
2015, CortexCitation Excerpt :Recognition and recall memory have been investigated more in PD than have recollection and familiarity, although several more selective studies of these two kinds of memory have been conducted. These studies have usually found that recollection is impaired, but that familiarity is preserved in mild to moderate nondementing PD (Algarabel, et al., 2010; Edelstyn et al., 2007; Edelstyn et al., 2010; Hay, Moscovitch, & Levine, 2002; RodrÍguez, Algarabel, & Escudero, 2014; Shepherd et al., 2013). However, familiarity has sometimes been found to be impaired as previously indicated (Davidson, et al., 2006; Weiermann, et al., 2010).
Subjective states associated with retrieval failures in Parkinson's disease
2013, Consciousness and CognitionCitation Excerpt :Hicks and Marsh (2002) and Souchay et al. (2007) noted that more accurate FOK judgements were associated with the subjective state of remembering (autonoetic) rather than Knowing (noetic). Using the Remember/Know paradigm most studies have found that recollection was spared in mild Parkinson’s (Davidson, Anaki, Saint-Cyr, Chow, & Moscovitch, 2006; Hay et al., 2002; but see Weiermann, Stephan, Kaelin-Lang, & Meier, 2010) and deficient in moderate PD (Edelstyn, Mayes, Condon, Tunnicliffe, & Ellis, 2007; Edelstyn, Shepherd, Mayes, Sherman, & Ellis, 2010; Hay et al., 2002; but see Weiermann et al., 2010). A disruption of recollection could offer an explanation for Episodic FOK impairments in this group.
The role of the basal ganglia in learning and memory: Insight from Parkinson's disease
2011, Neurobiology of Learning and Memory