Qualitatively different memory impairments across frontal lobe subgroups
Introduction
Damage to the frontal lobes does not result in the kind of severe amnesic syndrome typical of lesions to the temporal or diencephalic structures. However there is a large body of evidence suggesting that it may lead to a range of more subtle impairments of memory, particularly in recall tasks (Dimitrov et al., 1999; Janowsky, Shimamura, Kritchevsky, & Squire, 1989; Jetter, Poser, Freeman, & Markowitsch, 1986; Shimamura, Janowsky, & Squire, 1991; Wheeler, Stuss, & Tulving, 1995). Recall tasks are relatively effortful compared to recognition tasks, requiring the participant both to initiate an effective search in memory and to evaluate the products of this search. Therefore these deficits are thought to be secondary to impairments in frontally located supervisory processes, rather than being pure memory deficits.
Recall impairments in frontal lobe damaged patients have been attributed to difficulties in employing effective strategies at either encoding or retrieval. Several studies have reported that frontal patients tend not to spontaneously categorise to-be-remembered material or use other top-down processes to aid encoding (Incisa della Rochetta, 1986; Hirst & Volpe, 1988). Kopelman and Stanhope (1998) have also reported that the recall of frontal patients (in contrast to diencephalic and temporal lobe amnesics) can be improved if semantically organised rather than unrelated word lists are used, externally providing the organisation that they are unable to impose subjectively. Several other groups have reported similar results. Gershberg and Shimamura (1995) found that their frontal patients benefited from strategy instructions at both the study stage and at the test stage, implying that they had deficits in implementing organisational strategies at encoding and in implementing strategic processes at retrieval. Strategy deficits at retrieval in the form of impaired pair frequency have also been reported by Eslinger and Grattan (1994), Gershberg and Shimamura (1995) and Vilkki, Servo, and Surma-Aho (1998), and Incisa della Rochetta and Milner (1993) reported an improvement in recall when retrieval cues were externally provided, especially amongst left frontal patients.
Most lesion studies have used either an undifferentiated “frontal” group (Gershberg & Shimamura, 1995; Hirst & Volpe, 1988; Kopelman & Stanhope, 1998), or at best have compared left frontal to right frontal subgroups (Incisa della Rochetta, 1986; Incisa della Rochetta & Milner, 1993; Vilkki et al., 1998). However more anatomically specific results have emerged from groups using functional lesion localisation techniques to assess whether different deficits might follow lesions to different subregions of the frontal lobe. Stuss et al. (1994), for example, have reported that patients with left frontal (particularly left dorsolateral) damage had the most severe recall impairments in a list learning task. More recently Alexander, Stuss, and Fansabedian (2003) confirmed marked verbal free recall deficits in patients with posterior left dorsolateral lesions, but also in those with posterior medial frontal lesions, hypothesised to result from direct disruption of the memory system arising from loss of cholinergic projections to the hippocampus.
In addition to reduced veridical recall, patients with frontal lobe damage have frequently been reported to be abnormally sensitive to proactive interference, and to produce high rates of intrusions in recall tasks (Baldo, Delis, Kramer, & Shimamura, 2002; Daum & Mayes, 2000; Delbecq-Derouesne, Beauvois, & Shallice, 1990; Melo, Winocur, & Moscovitch, 1999; Shimamura, Jurica, Mangels, Gershberg, & Knight, 1995). However patient studies using more detailed lesion localisation procedures have failed to find any intrusion effects in their frontal groups (Alexander et al., 2003, Stuss et al., 1994) so the anatomical specificity of intrusion effects is not known. Stuss et al. (1994) did however report a specific deficit in patients with right lateral frontal lesions, who produced excess repetitions in their recall. This pattern of responding was attributed to an impairment in monitoring the output of recall, which prevented the patients from editing out words they had already recalled. Monitoring impairments have been associated with the production of false alarms in recognition memory tests (Budson et al., 2002; Curran, Schacter, Norman, & Gallucio, 1997; Delbecq-Derouesne et al., 1990, Melo et al., 1999; Schacter, Curran, Gallucio, & Milberg, 1996; Swick & Knight, 1999; Verfaillie, Rapscak, Keane, & Alexander, 2004) therefore right lateral monitoring impairments may also be related to the production of intrusions in recall.
Convergent with the patient findings, imaging studies of memory almost always show activation of the prefrontal cortex (PFC, see Fletcher & Henson, 2001, for a review). However for technical reasons related to movement artefacts, research in fMRI has tended to concentrate on recognition rather than recall. In general, greater left than right frontal activation is associated with encoding tasks, and greater right than left frontal activation is associated with retrieval tasks (the Hemispheric Encoding/Retrieval Asymmetry model: Habib, Nyberg, & Tulving, 2003; Nyberg, Cabeza, & Tulving, 1996; Shallice et al., 1994; Tulving, Kapur Craik, Moscovitch, & Houle, 1994).
Left prefrontal activation at encoding has most often been attributed to the retrieval of information from semantic memory which enables “deep” encoding to take place (Baker, Sanders, Maccotta, & Buckner, 2001; Henson, Rugg, Shallice, Josephs, & Dolan, 1999a; Wagner et al., 1998; Wig, Miller, Kingstone, & Kelley, 2004). More specifically this left prefrontal activation, and left dorsolateral activation in particular, has been associated with organisation of material at encoding on the basis of semantic relations or attributes (Fletcher, Shallice, & Dolan, 1998; Savage et al., 2001; Wagner, Maril, Bjork, & Schacter, 2001), a task at which patients with frontal lobe lesions are impaired.
Right prefrontal activation has been associated with retrieval success, or the adoption of a retrieval “mode” (Lepage, Ghaffar, Nyberg, & Tulving, 2000; Nyberg et al., 1996; Rugg, Fletcher, Frith, Frackowiak, & Dolan, 1997; Wagner, Desmond, Glover, & Gabrieli, 1998). More specifically, one component of this right prefrontal activation, and particularly right dorsolateral prefrontal activation, is hypothesised to reflect monitoring and checking of the products of a memory search (e.g. Cabeza, Locantore, & Anderson, 2003; Fletcher, Shallice, Frith, Frackowiak, & Dolan, 1996; Fletcher, Shallice, Frith, Frackowiak, & Dolan, 1998; Gabrieli, 1998; Henson, Shallice, & Dolan, 1999b; Shallice, 2001). In support of this, ERP studies have provided evidence of a late onsetting right frontal positivity in recognition tasks that follows ecphory, and this has been attributed to post-retrieval monitoring (see Allan, Wilding, & Rugg, 1998, for a review).
Shallice (2006) has assimilated this lesion and neuroimaging evidence in proposing that the Supervisory System (the prefrontal system which controls action in non-routine situations; Norman and Shallice, 1980, Norman and Shallice, 1986) may be fractionated into anatomically separable subsystems, including a left dorsolateral system which controls strategy production by means of top-down selection of schema (for example the semantic organisation of word lists), and a right dorsolateral system in charge of non-evident error-detection and checking (for example the monitoring functions discussed above). Shallice, 2001, Shallice, 2006 has specified the error detection, checking and monitoring functions of the right dorsolateral system by proposing that it comes into play under three conditions.
- (i)
When competing stimuli are likely to lead to capture errors. In memory paradigms, proactive interference provides the best test of the ability to reject plausible alternatives and avoid capture errors. Henson, Shallice, Josephs, and Dolan (2002), in an fMRI study, reported right dorsolateral PFC activation associated with high proactive interference in a paired associate cued recall paradigm. Despite evidence that frontal patients as a whole show abnormal sensitivity to proactive interference (Shimamura et al., 1995), the hypothesis that right frontal patients will be particularly sensitive to proactive interference and capture errors in a memory paradigm has not been directly tested.
- (ii)
When easy retrieval is over. The right dorsolateral error detection and checking system is proposed to come into play under conditions of uncertainty. For example, Henson, Rugg, et al. (1999) and Eldridge, Knowlton, Furmanski, Bookheimer, and Engel (2000) have both reported greater right dorsolateral PFC activation associated with “Know” rather than “Remember” responses, i.e. less certain responses that required greater checking activity. In a retrieval situation, we can assume that there are two stages: first the production of material which is easily available, and second a more effortful retrieval search for additional items which do not come easily to mind. The error detection and checking system would be most necessary at the second stage. Therefore patients with right lateral PFC damage might be expected to be more susceptible to capture errors in the later stages of retrieval.
- (iii)
When on-line monitoring is required. The non-evident error detection and checking system should be initiated especially when online monitoring is required. This is consistent with evidence that the mid-dorsolateral frontal cortex (areas 46/9) is involved in keeping track of actions and expected events so as not to repeat them (Petrides, 2000; Petrides, Alivisatos, Evans, & Meyer, 1993; Petrides, Alivisatos, Meyer, & Evans, 1993), and with the neuropsychological evidence cited above that patients with right frontal lesions produce abnormally high numbers of repetitions in their recall (Stuss et al., 1994). Therefore concurrent monitoring tasks should be particularly sensitive to right lateral PFC lesions.
A number of specific predictions deriving from theoretical and neuroimaging work can be made about the differential effects of lesions to different regions of the PFC. However few lesion studies thus far have been anatomically specific enough to test them. Most lesion studies have tended to use rather gross anatomical divisions, attributing impairments to a general “frontal” effect, and running the risk of masking more specific localised impairments. The present study employed more detailed lesion localisation methods in an attempt to localise specific recall impairments within the frontal lobe, and to close the gap between evidence from neuroimaging and lesion studies. Thirty-four patients with focal frontal lesions and 50 healthy controls were given a word list learning task, designed to assess four specific aspects of memory control.
- (i)
Proactive interference and intrusions. Lists were comprised of words drawn from four semantic categories. Semantic categories were repeated in subsequent lists, to induce proactive interference and the production of prior-list intrusions. The use of categorised lists might also induce the intrusion of unpresented semantic associates (extra-list intrusions). Previous research suggests that intrusion rates in general should be higher in patients with frontal lobe lesions (Baldo et al., 2002; Daum & Mayes, 2000; Delbecq-Derouesne et al., 1990, Melo et al., 1999, Shimamura et al., 1995). However Shallice, 2001, Shallice, 2006 error detection and checking hypothesis predicts that intrusion effects, and particularly proactive interference responses, should be highest in patients with right lateral PFC lesions.
- (ii)
The effects of prompting. After initial free recall, prompts in the form of category names were provided to prompt additional recall. If poor recall performance following frontal lesions is a result of impairments at retrieval, frontal patients should show an improvement in veridical recall following this manipulation (e.g. Gershberg & Shimamura, 1995; Incisa della Rochetta & Milner, 1993), and the HERA model suggests that this improvement should be most prominent in the Right Lateral group (Habib et al., 2003, Lepage et al., 2000, Nyberg et al., 1996, Shallice et al., 1994, Tulving et al., 1994). However failure of error detection and checking systems will also be most evident following prompting, as retrieval becomes non-routine and effortful, and this would predict an accompanying increase in intrusions, and particularly proactive interference responses, in the Right Lateral group.
- (iii)
Monitoring. A concurrent monitoring task was included in which participants were asked to indicate during recall, items that had already been presented in a previous list. Failures of on-line monitoring are hypothesised to result from failure of a right lateralised error-detection and checking system so should be higher in this group.
- (iv)
Organisational strategies at encoding and retrieval. An assessment of the use of organisational strategies was made by presenting word lists either blocked by semantic category, or randomly intermixed. Blocked presentation should aid recall, whereas random presentation requires subjective organisation on the part of the participants. On the basis of the HERA model any impairments in organisation at encoding should be particularly marked in the Left Lateral group (Fletcher, Shallice, & Dolan, 1998; Savage et al., 2001, Wagner et al., 2001), whereas impairments in organisation at retrieval should be particularly marked in the Right Lateral group (Habib et al., 2003, Lepage et al., 2000, Nyberg et al., 1996, Shallice et al., 1994, Tulving et al., 1994).
Section snippets
Participants
Thirty-four patients with focal frontal lesions were recruited from the National Hospital for Neurology and Neurosurgery and tested in the Neuropsychology Department. Inclusion and exclusion criteria were: (1) the presence of a focal lesion confined to the frontal lobes, (2) English as a first language, (3) absence of childhood onset epilepsy (late onset seizures arising from the lesion were allowed), (4) absence of severe aphasia, and (5) absence of other significant neurological and
Baseline neuropsychological testing
Baseline neuropsychological testing (see Table 2) showed that the Medial group had slightly depressed general intelligence as measured by Ravens APM performance compared to Controls (Univariate ANOVA: effect of “Group” p = 0.04, pairwise comparisons: Medial < Control p = 0.007). However there were no differences in NART estimated full scale IQ, and no naming or visual perception impairments in any group.
Free recall
The free recall performance of the Control and Frontal groups on the experimental task can be seen
Discussion
Our study confirms previous reports that patients with frontal lobe lesions may show impairments on verbal free recall tasks (Dimitrov et al., 1999, Janowsky et al., 1989, Jetter et al., 1986, Shimamura et al., 1991, Wheeler et al., 1995). However the use of more detailed anatomical groupings and the testing of specific predictions from neuroimaging and recent theoretical work have revealed more detailed results than previously reported. The first striking finding was that it was only the
Acknowledgements
We would like to thank Miss Joan Grieve, Dr Neil Kitchen, Mr Michael Powell, Dr D.G. Thomas and Mr Laurence Watkins for permission to study the cognitive performance of patients under their care, and Ms Bonnie-Kate Dewar who assisted in testing of some of the patients. Martha Turner is supported by an ESRC/MRC Postdoctoral Fellowship PTA037270085.
References (64)
- et al.
Electrophysiological evidence for dissociable processes contributing to recollection
Acta Psychologica
(1998) - et al.
False recognition after a right frontal lobe infarction: Memory for general and specific information
Neuropsychologia
(1997) - et al.
Associative learning impairments in patients with frontal lobe damage
Brain and Cognition
(1999) - et al.
Altered serial position learning after frontal lobe lesion
Neuropsychologia
(1994) - et al.
Impaired use of organizational strategies in free recall following frontal lobe damage
Neuropsychologia
(1995) - et al.
Hemispheric asymmetries of memory: The HERA model revisited
Trends in Cognitive Sciences
(2003) - et al.
Functional magnetic resonance imaging of proactive interference during spoken cued recall
NeuroImage
(2002) - et al.
Memory strategies with brain damage
Brain and Cognition
(1988) - et al.
A verbal long-term memory deficit in frontal lobe damaged patients
Cortex
(1986) - et al.
Recall and recognition memory in patients with focal frontal, temporal lobe and diencephalic lesions
Neuropsychologia
(1998)