Some older adults without noticeable cognitive problems have a harder time than younger people in separating irrelevant information from what they need to know at a given time, and a new Johns Hopkins University study could explain why.
The findings offer an initial snapshot of what happens in the brain as young and old people try to access long-term memories, and could shed light on why some people’s cognitive abilities decline with age while others remain sharp.
“Your task performance can be impaired not just because you can’t remember, but because you can’t suppress other memories that are irrelevant,” said senior author Susan Courtney, a cognitive neuroscientist at Johns Hopkins.
“Some ‘memory problems’ aren’t a matter of memory specifically, but a matter of retrieving the correct information at the right time to solve the problem at hand.”
The findings were just posted in Neurobiology of Aging.
The researchers had 34 young adults (18 to 30) and 34 older adults (65-85) perform a mental arithmetic task while their brain activity was measured through functional magnetic resonance imaging, or fMRI.
Other images were also collected to measure the integrity of the connections between brain areas called white matter tracts.
The task compared the participants’ ability to inhibit irrelevant information automatically retrieved from long term memory.
They were asked to indicate whether a proposed solution to an addition or multiplication problem was correct or not—for instance 8×4=12 or 8+4=32.
These examples would create interference as participants considered the right answer because although they should answer “incorrect,” the proposed solution seems correct at first glance, based on long-term memories of basic math.
This interference did not exist when participants were asked to answer clearly false equations like 8×4=22.
Making the task even more complicated, the subjects were sometimes asked to switch to multiplication after they saw the addition symbol and vice versa.
Older people were a fraction of a second slower at answering the questions than younger participants, particularly when there was interference, but the more dramatic difference showed up in the brain scans.
Older individuals who had more difficulty with interference also had more frontal brain activation than young adults.
The brain imaging demonstrated that in some aging participants, fibers connecting the front and back of the brain appear to have been damaged over the years.
However other older individuals had fibers similar to much younger subjects.
The greater the integrity of these fibers, the better the participant’s task performance, said lead author Thomas Hinault, a postdoctoral fellow at Johns Hopkins.
“Everyone we studied had good functioning memory, but still we saw differences,” Hinault said.
“There are so many disruptions in the world and being able to suppress them is crucial for daily life.”
The researchers were surprised to find that during parts of the task that were the trickiest, where participants had to switch between multiplication and addition and were asked to add after they saw a multiplication command or vice versa, the people with the strongest brain fiber connections counterintuitively performed even better.
Something about deliberately exercising the mind in this fashion made the most agile minds even more so.
“If you have good connections between brain networks, that will help,” Courtney said. “If not, you have interference.”
Semantic cognition is supported by two interactive components: semantic representations and mechanisms that regulate retrieval (cf. ‘semantic control’).
Neuropsychological studies have revealed a clear dissociation between semantic and episodic memory.
This study explores if the same dissociation holds for control processes that act on episodic and semantic memory, or whether both types of long-term memory are supported by the same executive mechanisms. We addressed this question in a case-series of semantic aphasic patients who had difficulty retrieving both verbal and non-verbal conceptual information in an appropriate fashion following infarcts to left inferior frontal gyrus (LIFG).
We observed parallel deficits in semantic and episodic memory: (i) the patients’ difficulties extended beyond verbal materials to include picture tasks in both domains; (ii) both types of retrieval benefitted from cues designed to reduce the need for internal constraint; (iii) there was little impairment of both semantic and episodic tasks when control demands were minimised; (iv) there were similar effects of distractors across tasks.
Episodic retrieval was highly susceptible to false memories elicited by semantically-related distractors, and confidence was inappropriately high in these circumstances.
Semantic judgements were also prone to contamination from recent events.
These findings demonstrate that patients with deregulated semantic cognition have comparable deficits in episodic retrieval.
The results are consistent with a role for LIFG in resolving competition within both episodic and semantic memory, and also in biasing cognition towards task-relevant memory stores when episodic and semantic representations do not promote the same response.
Neuropsychological studies provide compelling evidence for the existence of separable episodic and semantic memory stores.
Patients with semantic dementia have progressive yet selective degeneration of conceptual knowledge across all tasks and input modalities, which correlates with the degree of atrophy in the anterior ventrolateral temporal lobes (Butler et al., 2009, Mummery et al., 2000), yet their memory for recent episodic events is largely intact (Graham and Hodges, 1997, Graham et al., 1997, Graham et al., 2003, Graham et al., 2000).
In contrast, anterograde amnesia is characterised by poor encoding and retrieval of specific events as opposed to factual information, following damage to the hippocampus and associated structures in the medial temporal lobes (Nadel and Moscovitch, 1997, Nestor et al., 2006, Vargha-Khadem et al., 1997). These findings suggest that anterior ventrolateral temporal cortex supports conceptual generalisation across experiences, while hippocampus promotes pattern separation for recently-encoded episodes (Kumaran and McClelland, 2012, McClelland et al., 1995).
Studies also point to the existence of contrastive types of semantic deficit.
The term “semantic aphasia” was first coined by Head (1926) to describe patients showing difficulties in shaping and manipulating knowledge to serve symbolic processing – in the presence of heterogenous language impairments – rather than loss of semantic knowledge per se.
In line with Head’s clinical description, studies have shown that, unlike the degraded knowledge in semantic dementia, patients with semantic aphasia (SA) show deregulated semantic cognition across different tasks and input modalities following left frontoparietal stroke (Jefferies and Lambon Ralph, 2006, Jefferies et al., 2008, Rogers et al., 2015).
SA patients show inconsistent semantic performance when the same concepts are tested under different control demands, as well as sensitivity to cues and miscues that constrain retrieval or increase the availability of irrelevant knowledge (Corbett et al., 2011, Jefferies et al., 2008, Noonan et al., 2010).
They have difficulty retrieving non-dominant aspects of knowledge and dealing with competition from strong yet irrelevant semantic distractors during semantic retrieval (Almaghyuli et al., 2012, Noonan et al., 2010). These problems extend beyond language, to affect sound, picture and action understanding (Corbett et al., 2009a, Corbett et al., 2009b, Corbett et al., 2011, Gardner et al., 2012, Thompson et al., 2015).
Collectively this evidence shows that SA patients have multimodal deficits of semantic control, i.e., they find it difficult to flexibly retrieve and shape semantic knowledge to suit the task or circumstances and show impairment when there is a need to resolve competition between different meanings or features of concepts.
The distinction between semantic dementia and patients with SA supports a component process account, in which semantic cognition emerges from interactions between transmodal conceptual representations and control processes (Controlled Semantic Cognition Framework; Jefferies, 2013, Lambon Ralph et al., 2017).
This proposal is also pertinent to understanding differences in episodic memory deficits in amnesia (see Blumenfeld & Ranganath, 2007 for a review).
In contrast to patients with circumscribed medial temporal lobe injury (such as HM, Scoville & Milner, 1957), patients with additional prefrontal involvement show better cued than free recall (Mangels et al., 1996, della Rocchetta and Milner, 1993) and disproportionate difficulty in retrieving word-pairs previously associated with other targets, reflecting a failure to overcome proactive interference (Shimamura, Jurica, Mangels, Gershberg, & Knight, 1995).
In both semantic and episodic tasks, bringing to mind unusual associations, or task-relevant knowledge in the face of strong competition, might involve promoting specific aspects representations and suppressing irrelevant dominant information (Anderson, 1988; Badre and Wagner, 2007, Whitney et al., 2011).
The similarity of these theoretical accounts fuels interest in whether they have a shared or distinct neural basis.
Functional neuroimaging studies suggest that overlapping networks are important for the control of episodic and semantic memory (see Fig. 1A).
Left inferior frontal gyrus (LIFG) has a well-established role in the control of episodic memory: it shows a stronger response in the retrieval of weakly vs. strongly-encoded memories (Barredo et al., 2015, Hayes et al., 2011) and is engaged by interference resolution (Badre and Wagner, 2005, Wimber et al., 2009).
Likewise, this region shows increased activation in semantic retrieval for ambiguous words, weak associations or strong distracters (for a meta-analysis, see Noonan, Jefferies, Visser, & Lambon Ralph, 2013; also Badre and Wagner, 2005, Badre and Wagner, 2007, Thompson-Schill et al., 1997).
Controlled retrieval from episodic and semantic memory partially overlaps with “multiple-demand regions” that are engaged for difficult tasks across multiple domains; however, anterior LIFG lies outside this network and appears to specifically support the control of memory (Badre et al., 2005, Davey et al., 2016, Nelson et al., 2009).
In line with this proposal, inhibitory transcranial magnetic stimulation to LIFG disrupts control-demanding semantic judgements but not more automatic aspects of semantic retrieval or demanding non-semantic judgements (Gough et al., 2005, Hallam et al., 2016, Hoffman et al., 2010, Krieger-Redwood and Jefferies, 2014, Whitney et al., 2011).
Brain networks implicated in semantic and episodic retrieval overlap with patients’ lesions. (A) Semantic control network (red, from Noonan et al., 2013, adapted by Humphreys and Lambon Ralph, 2015), episodic memory network (green, from Neurosynth; a meta-analysis of 393 studies containing the term “episodic”), the overlap of the two networks (yellow). Rendered views are displayed using Surfice (https://www.nitrc.org/projects/surfice/); sagittal views using MRIcroGL (http://www.cabiatl.com/mricrogl/). The overlap mask included only one cluster of a minimum of 50 voxels which corresponded to mid-to-post LIFG, pars triangularis extending to pars opercularis and middle frontal gyrus (MNI -48, 24, 24). (B) Lesion overlay of the sample of SA patients included in the study. Patients’ brains compared to aged-matched controls. Grey matter, white matter and CSF were segmented and changes from the healthy control brains were highlighted as ‘lesion’ using automated methods (Seghier et al., 2008). Colour bar indicates amount of overlap from 1 to 10 patients.
Despite these similarities, few studies have directly compared manipulations of difficulty across episodic and semantic judgements.
It is unclear whether LIFG contributes to episodic memory indirectly by regulating conceptual retrieval or whether LIFG is crucial for regulating retrieval from both memory stores.
Neuropsychology can help to resolve this theoretical uncertainty by establishing if damage to LIFG gives rise to symmetrical deficits of episodic and semantic memory. Semantic and episodic representations often mutually support retrieval: to understand the semantic link between items like dog and beach, we can bring to mind specific episodes in which these items co-occurred (Westmacott and Moscovitch, 2003, Westmacott et al., 2004).
Similarly, in event memory, we draw on semantic representations of related episodes to support encoding and retrieval, giving rise to “levels of processing effects” (Anderson, 1981, DeWitt et al., 2012). We therefore need the capacity to select a response from one or other system, depending on the task demands.
The inappropriate application of semantic information in an episodic context can give rise to false memories (Roediger, Balota, & Watson, 2001; Roediger & McDermott, 1995) and the engagement of LIFG might help to avoid these errors (Dennis et al., 2014, Garoff-Eaton et al., 2007, Kim and Cabeza, 2007).
In this study, we examined chronic post-stroke patients with SA and well-documented deficits of semantic control following LIFG lesions.
To date, there has been little research on episodic memory in aphasia, including SA. We therefore investigated whether SA patients would show episodic deficits resembling their semantic impairment – namely, multimodal difficulties across verbal and non-verbal tasks, and sensitivity to cues that reduce the requirement for internally-constrained retrieval. We assessed whether semantic control impairment would elicit ‘false episodic memories’.
In addition, to establish if semantic deficits directly underpin poor episodic memory or, alternatively, whether LIFG is critical for memory control across domains, we considered whether LIFG lesions would elicit ‘false semantic associations’ when semantic retrieval is preceded by task-irrelevant episodic encoding.
Patients with multimodal semantic deficits following infarcts within LIFG may have difficulty resolving competition between episodic and semantic memory and their responses might reflect task-irrelevant memory representations, if LIFG plays a general role in regulating retrieval from both systems.
More information: Thomas Hinault et al, Age-related differences in the structural and effective connectivity of cognitive control: a combined fMRI and DTI study of mental arithmetic, Neurobiology of Aging (2019).DOI: 10.1016/j.neurobiolaging.2019.06.013
Journal information: Neurobiology of Aging
Provided by Johns Hopkins University
