Apathy and Actions- Another Consideration when Theorizing about Embodied Nature of Language in Parkinson’s Disease

Abstract

Objectives:

Individuals with Parkinson’s disease (PD) have difficulty processing actions or verbs relative to nouns. Verb difficulties are thought to represent the coupling of language and motor networks. However, those with PD also frequently experience apathy. The overlap between apathy and action language difficulties is to date unexplored. Furthermore, whether verb/action fluency difficulty represents verb degradation (semantic/conceptual) or a selective lexical retrieval difficulty has not been determined.

Methods:

In the current study, 20 individuals with PD without dementia completed cued action (verb) and animal (noun) fluency tasks in addition to assessments of apathy, cognition, and motor functioning.

Results:

Individuals who exhibited impairments on action and animal fluency improved around 50% with the provision of cueing. The degree to which action fluency improved with cueing was correlated with behavioral/initiation apathy (r_s=.56) as well as motor dysfunction (r_s=−.57), while no similar relationship was found between those factors and nouns.

Conclusions:

These findings suggest that impaired retrieval of actions and nouns are present in PD, but may have different underlying neuropsychological underpinnings. This provides preliminary support for grounded cognition models, which suggest the brain organizes information around motor, perceptual, and other networks.

Introduction

Parkinson’s Disease (PD) is more than a disorder of movement, as deficits are exceedingly common in cognition, behavior, and other non-motor domains (Lee & Koh, 2015). Language difficulties, such as word retrieval problems, are frequently reported by individuals with PD (Raein et al, 2019) and their caregivers (Wolff & Benge, 2019) and are commonly observed on measures such as generative verbal fluency (Henry & Crawford, 2004). This difficulty may be particularly observed for action/verb language tasks. The reason for this pattern of impairment, and the underlying nature of language impairment in PD remains a relatively unexplored area.

Action Language Difficulties in PD and their Neurological Substrates

In recent years, evidence has emerged suggesting a disproportionate impairment in verb (action) relative to noun processing in individuals with Parkinson’s disease (PD). This action weaker than noun finding can be found early in the disease process and seems dissociable from general cognitive decline (Bocanegra et al, 2015). These deficits are demonstrated on a variety of tasks, such as verb generation to a particular cue (Péran et al., 2009), action fluency (McDowd et al., 2011), and even basic action naming tasks (Boulenger et al., 2008). The co-occurrence of verb processing difficulties with a primary motor disorder has been thought to provide evidence of “embodied” or “grounded” cognition, wherein higher order cognitive processes are grounded in lower level sensory and motor processes (Barsalou, 2008; Barsalou, 2010; Damasio, 1989; Salmazo-Silva et al, 2017). Embodied cognition models suggest that the brain uses more primary systems to store complex concepts rather than extracting and storing such conceptual information in some independent coded representational manner. In this respect, modern theories of the substrates of representation of language, such as Martin’s (2016) GRAPES model (Grounding Representations in Action, Perception, and Emotion Systems), explicitly postulate that language representations are grounded in a host of systems, including neuropsychiatric substrates.

In a disease such as PD, where neuropsychiatric as well as motor systems can be affected, embodied cognition models (such as GRAPES) may represent a conceptual framework for understanding critical neural correlates of action language functioning. Neuropsychiatric problems are common in PD and include depression (e.g., Cummings, 1992), anxiety (e.g., Dissanayaka et al., 2016), impulse control disorders (e.g., Weintraub et al., 2010), and apathy. Apathy is defined as reduced motivation towards goal directed behaviors (Marin, 1996). By definition, apathy is a condition that cannot be attributable to decreased level of consciousness, cognitive impairment or emotional distress, and is rather a state of primary motivational impairment (Marin, 1990). Apathy correlates with motor dysfunction and depression in pwPD, though only modestly, and frequently emerges as a distinct and difficult to treat neuropsychiatric symptom independent from (and in the absence of) depression (Cubo et al, 2012).

Three underlying dimensions of apathy have been demonstrated in the literature and Parkinson’s disease (Radakovic et al., 2018): cognitive/executive, emotional-affective, and auto-activation (Levy & Dubois, 2006). Cognitive/executive apathy is described as an inability to strategize plans and actions to achieve a goal. In contrast, emotional/affective apathy refers to the marked absence of emotional drive to activity. Finally, there is behavioral initiation (also known as uto activation), which represents the reduced initiation of thoughts and behavior related to daily activities.

Apathy has been demonstrated to be a common difficulty in persons with Parkinson’s disease (pwPD), affecting between approximately 30 to 50% of all pwPD (Duardin et al., 2014; Kirsh-Darrow et al., 2006; Pedersen et al., 2000; den Brok et al., 2015). Though apathy and action language deficits are both frequently observed in individuals with PD, there has been relatively little work exploring any potential relationship between these symptoms.

When viewed through the lens of the GRAPES Model, the question emerges: could an emotional or motivational system that reduces action initiation (apathy), in addition to motoric dysfunction, serve as a substrate for action language disruption? In other words, do individuals in a state where they seem unable to generate actions in the real world also show a concomitant difficulty in generating action language? Data are currently lacking in this area, yet there are hints that an apathy/language overlap may exist. For example, individuals with PD and apathy performed significantly worse than individuals with PD but no apathy on both phonemic and semantic generative verbal fluency tasks (Zgaljardic et al., 2007). Similarly, Isella and colleagues (2002) divided pwPD into three groups based on their scores on an apathy scale, and found that those who were most apathetic were significantly more impaired with letter and category fluency tests. However, none of these studies have explored directly the relationship between verb processing and apathy in particular, but have instead focused on the general idea that persons who suffer from an amotivational syndrome will have reduced behavioral output across the board. However, to date no studies in our knowledge have tested the idea that verb or action generation tasks in particular may show dissociable patterns of association with apathy. Further, as we address next, there may be differences in both substrate and mechanism of retrieval difficulties that need to be considered in individuals with PD.

Action Language Difficulties in PD: A Difficulty with Retrieval or Semantic Degradation?

While semantic fluency impairments are common in both PD and other conditions such as Alzheimer’s disease (Chasles et al, 2019; Lim et al, 2019) the underlying mechanism of impairment appears to differ by disease and neurological substrate. For example, cued fluency trials, where more specific semantic cues are administered after an initial word generation period have demonstrated different results between groups. Individuals with PD and Huntington’s disease significantly benefited from cueing on a semantic fluency task when compared to those with Alzheimer’s disease (Randolph et al, 1993), who did not improve. These findings were thought to represent a difficulty in retrieval in fronto-subcortical disorders such as PD, but a problem with underlying semantic degradation (i.e., the concepts are gone) in individuals with Alzheimer’s disease. Similarly, Drane and colleagues (2006) examined cued semantic fluency performance of individuals with complex partial seizures of frontal or temporal lobe onset. Those with frontal lobe seizures showed significantly greater performance improvement when provided with a cued semantic fluency format compared to those with temporal lobe epilepsy (TLE). In subsequent work, Drane and colleagues (2013) have proffered the idea that rather than a semantic degradation of information, the failure to benefit from cueing on semantic fluency tasks and visual confrontation naming deficits observed in those with TLE are likely resulting from an access problem. They postulate that information is likely stored diffusely and not degraded but cannot be accessed due to disruptions of a primary mechanism of lexical retrieval not counteracted by cueing. The lexical retrieval problem in TLE would be driven by a “decoupling” of expressive language functions from regions dedicated to semantic storage. In contrast, patient’s with fronto-subcortical diseases would exhibit lexical retrieval problems that are based on underlying executive dysfunction (e.g. poor organization and planning, inability to shift between subcategories; Galtier et al., 2017). To date, no studies have examined whether action fluency difficulties in PD may represent degraded semantic store for action words or a primary or a lexical retrieval-based difficulty (which in turn may be due to underlying executive dysfunction).

Specific Aims

While it is known that verbal fluency (e.g., letter, semantic, and verb) can be decreased in pwPD, very little is known about the mechanistic and conceptual underpinnings of these deficits. In the current preliminary study, we hope to improve this knowledge by proposing the following aims which seek to address potential differential substrates of verbal fluency tasks as well as better understand mechanisms of difficulty.

Aim 1: The Relationship Between Cued Verb and Noun Generation, Motor Performance, and Neuropsychiatric Features:

We hypothesize that action fluency performance will be strongly correlated with measures of motor dysfunction and apathy but not with measures of mood more generally (depression, anxiety) in pwPD. Moreover, we hypothesize that there will be a significant inverse correlation between motor performance/apathy and benefit from verb cueing, such that patients with the greatest motor difficulties in these domains benefit the most from cueing. These patterns will not be seen for common noun fluency. Combined, these findings would be consistent with the grounded cognition models of information processing (e.g., GRAPES), and would provide a rationale for exploring these models with neuroimaging and broader experimental paradigms in future studies.

Aim 2: Clarifying Whether Action and Semantic Fluency Difficulties Represent Lexical Retrieval Difficulties or Degradation of Semantic Information:

We hypothesize that pwPD will show significant benefit with cueing for both action and semantic fluency tasks, as we expect that limitations on this task are primarily due to retrieval difficulties which will respond to a cued paradigm. If one or both categories do not respond to cueing, then it is likely that the underlying cognitive mechanism of dysfunction involves either semantic degradation or diminished lexical access not based on executive retrieval/control deficits.

Methods

Institutional Review and Participant Safety

All study procedures were reviewed and approved by our local institutional review board. Participants provided written informed consent prior to participating.

Participants

Twenty participants were recruited from our local movement disorder center via clinician referral and from movement disorder support groups in our catchment area. Interested individuals were screened for inclusion based upon record review and phone interview with a trained research assistant. Individuals were required to have a PD diagnosis made by a neurologist, which was verified with referring clinician, chart review, and/or structured clinical interview with the pwPD to rule out symptoms concerning for an atypical parkinsonism. Individuals were excluded from participation if they had other conditions which could meaningfully impact cognition, including any history of traumatic brain injury with loss of consciousness greater than 30 minutes, prior stroke, diagnosis of dementia, prior brain surgery (including deep brain stimulator implantation), prior diagnosis of encephalitis, active psychotic symptoms, current alcohol or illicit drug abuse, or history of severe mental illness. Because of the language focused nature of this project, we excluded individuals who reported a history of language learning disability, non-native English speakers, and those with hearing or vision problems that could interfere with testing.

For the current study, we focused on individuals without dementia (individuals with MCI were included) to avoid the potential confounding effect of global cognitive impairments on measures of interest. To this end, potential participants were first screened with a variant of the telephone version of the Montreal Cognitive Assessment (MoCA; Wong et al., 2015; Zietemann, et al., 2017). Our version consisted of 21 points and, we used a cut point of 15 to exclude individuals from participation in our study due to dementia (Benge & Kiselica, 2021). After completion of all study activities described below, participants were given a voucher worth $100 US.

Neuropsychological Evaluation and Cognitive Categorization

Participants completed all assessment activities in a half-day session that typically lasted 2.5 to 4 hours. They were tested on their usual medication regimen in the “on” state. All assessments were conducted by a trained neuropsychology post-doctoral fellow, intern, or neuropsychologist. To determine the pwPD’s overall cognitive status, a neuropsychological battery with two tests per domain (consistent with Level II assessment for mild cognitive impairment per Movement Disorder Society recommendations; Litvan et al., 2012) was administered. Verbal Memory was assessed with the Rey Auditory Verbal Learning Test (RAVLT; Schmidt, 1996). Visual Memory was assessed by either the Brief Visuospatial Memory test (BVMT-R; Benedict et al., 1996) or Brown Location Test (BLT; Brown et al., 2007); 4 individuals were missing a visual memory measure. Attention and Working Memory were assessed by digit span forward and Trailmaking Part A, with versions used from the Uniform Data Set 3.0 neuropsychological battery (Weintraub et al., 2009). Executive functioning assessment included the CLOX clock drawing (Royall et al., 1998) and performance on the Trailmaking Test Part B [TMTB]). Visual skills were assessed by performance on the short form of the Judgment of Line Orientation test (JLO; Benton et al., 1994; Woodward et al., 1996) and the short form of the Benton Facial Recognition Test (BFRT; Benton et al., 1983; Levin et al., 1975). To define the core language domain for this study, performance on the Boston Naming Test (BNT; Kaplan et al., 2001) and semantic fluency for animals were utilized. Scores <=−1.5 standard deviations below the normative mean were considered impaired. In keeping with MDS Level II guidelines (Litvan et al., 2012), individuals were categorized as having mild cognitive impairment (MCI) if more than two tests within a single cognitive domain or two tests across the battery above were impaired.

Generative Verbal Fluency Measures

In addition to the core neuropsychological measures, several experimental language and cognitive measures were administered. Most pertinent to this study, we included the administration of cued semantic(animal) and verb (action) fluency measures. For conceptual clarity in this manuscript we will discuss more common category fluency tasks such as animal fluency as an exemplar of semantic fluency, distinguishing this from action or verb fluency which for reasons outlined in the introduction is thought to represent an important subclass of language in individuals with PD. Cued generative verbal fluency measures (Randolph et al., 1993, Drane et al., 2006) are similar to more traditional verbal fluency measures in their initial trial administration. For animal fluency, individuals were first prompted to name as many animals as they could in 60 seconds. Next, individuals were given prompts to name as many animals as they could that were pets, found on the farm, lived in the jungle, or lived in the water. Fifteen seconds were given for each of these subcategory prompts. Drane and colleagues have extended this paradigm to a number of semantic and proper noun fluency tasks, based upon the growing recognition that these constructs are differentially represented at a neural level, with various temporal lobe regions more associated with certain item types than others (e.g. the temporal pole has been implicated in processing unique entities; Waldron et al., 2014; Drane & Pedersen, 2019).

Action/Verb fluency was assessed using a similar 60 second initial prompt using task instructions from Piatt et al. (1999). A cued version of the test was also created to parallel the animal fluency version. Briefly, after the initial 60 second trial, participants were given 15 seconds to generate actions they would do with their hands, legs, minds, and mouths. The sum of cued responses across the 4 fifteen-second cued trials (which could include repeated words from the initial trials) was calculated, and divided by the number of uncued (original 60 second fluency trials). This provided a variable that represented the percentage of improvement in performance with cueing. Cued verbal fluency measures were administered prior to the full neuropsychological battery, and administered in a random order to each participant.

Motor and Functional Status Measures

Overall Parkinson’s motor and disability staging was completed by study staff using Hoehn and Yahr rating (H&Y; Hoehn & Yahr, 1967). The H&Y is a clinical rating that combines degree of motor impairment with functional difficulties, ranging from a rating of I (unilateral motor symptoms but independent for ambulation and activities) to V (wheelchair bound or bed ridden unless assisted). Speeded fine motor dexterity was also assessed by performance on the 9-Hole Peg Test (9-HPT; Wang et al.,, 2015).

Neuropsychiatric Measures.

Participants were administered the self-reported Dimensional Apathy Scale (DAS; Radakovic and Abrahams, 2014). The DAS assesses 24 items in three empirically derived subscales that reflect cognitive/executive, emotional/affective, and behavioral initiation apathy. Participants utilize a 4-point scale (i.e., almost always, often, occasionally, hardly ever) to rate the way they felt, behaved, or thought over the last month. Each domain has eight questions, for a maximum total score of 72 (24 points in each domain). The DAS has been evaluated in PD, and showed high internal consistency and good evidence for convergent and discriminant validity (Santangelo et al., 2017). For the current study, DAS scores were transformed into Z-scores on the basis of normative data (Radakovic et al, 2016).

Participants also completed short forms of anxiety and depression questionnaires, from the NEURO QOL system (Cella et al., 2012), a system of self-report measures developed explicitly for use in individuals with neurodegenerative disorders. Participants used a 5-point scale (i.e., never, rarely, sometimes, often, always) to rate anxiety and depression symptoms over the past seven days. There are eight questions on each measure, for a maximum total of 40 points.

Statistical Analyses

To address questions related to our first aim, we evaluated the relationship between cued and uncued action and semantic fluency tasks with measures of motor functioning, apathy and measures of depression and anxiety. We hypothesized that action language difficulties in particular would correlate with apathy in addition to motor dysfunction, suggesting a shared substrate of action language, apathy (action generation) and motor functions. In contrast, such relationships should not be seen with depression and anxiety. Spearman correlations were used to address the relationship among variables, with correlations <.4 considered weak, .4–6 considered moderate, and .6 or above considered strong.

Our second aim sought to evaluate whether action and noun fluency impairments were more related to a lexical retrieval problem versus a lexical access problem due to semantic access/degradation. As with prior studies, we assume that a failure to benefit from cueing would reflect either a lexical access problem or to a problem with semantic access/degradation. We contrasted these mechanistic explanations with two analyses. First, we compared the extent that performance improved with cues for animal and action fluency with the sample as a whole with paired sample t-tests. Next, we explored if those with impaired performances on the typical (60 second) animal and action fluency trials benefitted more with cueing than those without impairments. Given the exploratory nature of this study, we did not correct for multiple comparisons and considered relationships with p values <.05 to be significant and trends p <.1.

Results

Descriptive and Clinical Characterization of the Sample

Twenty persons with PD were included in the current study. Demographic, motor, and cognitive classification data for the sample are presented in Table 1, with descriptive data for the neuropsychological variables in Table 2. Descriptive data for the measures of core interest to the study (action/verb and animal fluency, self-report measures, and fine motor dexterity) are reported in table 3.

Table 1-.

Demographic and clinical characterization of the sample (n=20)

Variable
Age (years)	72.25 (6.36)
Education	Some College: 5% Associate’s Degree: 5% Bachelor’s Degree: 60% Master’s Degree: 10% Doctorate Degree: 20%
Gender	Male: 55% Female: 45%
Ethnicity	Caucasian: 100%
Hoehn & Yahr (H&Y) Stage	I: 85% II: 5% III: 10%
Cognitive Diagnosis	No Cognitive Diagnosis: 80% Mild Cognitive Impairment: 20%

Table 2:

Demographically Corrected Neuropsychological Scores in the Sample (n=20)

Variable	Mean T or Median Percentile	SD or IQR
Verbal Delayed Recall	51.55	11.45
Visual Delayed Recall	34.78	23.42
Trailmaking A	50.75	8.50
Forward Digit Span	48.50	8.89
Boston Naming Test	57.63	9.12
Trailmaking B	49.16	8.02
Clock Drawing	42.17	17.5
Judgment of LineOrientation	68.0%	46%
Facial Recognition	47.50	11.05

Note: All scores are normatively corrected as outlined in the methods section; SD = standard deviation; IQR = Interquartile Range

Table 3:

Motor, Emotional/Behavioral Functioning, and Action, and Animal Fluency Descriptive Data for the sample (n=20)

Variable		Mean	SD
Dimensional Apathy Scale (All scales with maximum values of 24)	Executive/Cognitive Subscale	7.65	3.59
	Emotional/Affective Subscale	9.85	2.60
	Behavioral Initiation Subscale	11.05	4.38
Anxiety		16.80	5.18
Depression		13.50	6.73
Action Fluency	60 Second Trial	14.05	5.44
15 Second Action Cued Fluency Trials	Arm	4.00	1.34
	Leg	4.55	3.80
	Mind	3.80	0.95
	Mouth	4.95	1.57
	% Improvement with Cues	29.06	40.73
Animal Fluency	60 Second Trial	19.15	5.89
15 Second Animal Cued Fluency Trials	Pets	5.79	1.58
	Farm	6.16	1.61
	Jungle	5.39	2.09
	Water	4.94	1.47
	% Improvement with Cues	20.32	26.07
9 Hole Peg (T score)	Dominant Hand	43.63	10.37
	Non- Dominant	43.70	16.49

Aim 1 Analyses: The Relationship Between Action and Noun Generation, Motor Performance, and Neuropsychiatric Features

The relationship (Spearman correlations) between dexterity, apathy dimensions, mood and generative verbal fluency performances are presented in Table 4. Action fluency performance correlated modestly (r_s = −.43, p=.06) only with self-reported behavioral/initiation apathy dimension. In other words, increasing self-reported inability to initiate actions was associated with fewer spontaneously produced verbs. However, improvement in action generation with cues was positively associated with behavioral initiation apathy with a strength of relationship considered moderate (r_s= .56, p<.01). That is, individuals who endorse behavioral initiation apathy also seem to be the ones who benefit the most when provided with a cue or structure. No other dimension of apathy, nor self-reported anxiety or depression demonstrated this relationship.

Table 4:

Spearman Correlations Between Action and Semantic Fluency, Apathy, Mood, and Motor Performances

	Behavioral Initiation Apathy	Cognitive/Executive Apathy	Emotional/Affective Apathy	Depression	Anxiety	9 Hole Peg -Dominant	9 Hole Peg- Non Dominant
Action Fluency Total	−.43a	−.18	−.24	−.33	−.19	.49*	.35
% Improvement With Action Cues	.56**	.27	.28	.36	.24	−.57**	−.35
Animal Fluency Total	.12	.29	−.23	.11	−.10	.15	.19
% Improvement with Animal Cues	−.17	−.15	.39a	−.09	−.14	−.08	−.07

Note:

^a=trend (p<.10);

^*p < .05,

^**p < .01,

Dom. = Dominant hand; NonDom = Non-Dominant

Speeded fine motor dexterity correlated positively with action fluency performance (r_s= .49, p=.01) with faster pegboard performance associated with better performance. An inverse and moderate relationship was observed between percentage improvement with cues and motor speed (i.e. slower motoric performance was associated with greater benefit from cues (r_s = −.57, p<.01).

The same pattern was not found when animal fluency was considered. While a trend was noted (r_s=.39, p=0.10) between animal fluency and percent improvement with cues and emotional apathy, there was otherwise no relationship observed between apathy, motor functioning, and animal fluency performance.

Aim 2 Analyses: Clarifying whether Action and Semantic Fluency Difficulties represent Retrieval Difficulties or Degradation of Semantic Information

The provision of cues significantly improved both action (t(19)=−2.96, p<.01) and animal fluency (t(18)=−3.65, p<.01). The overall percentage of improvement with cues did not differ significantly between animal and action categories (t(18)=0.96, p=.35.). Overall, these findings would suggest a similar mechanism (impaired retrieval not degradation/decreased access to semantic concepts per se) for both actions and nouns in in persons with PD.

We also evaluated if those with clinically significant verbal fluency impairments showed improvement with cueing across tasks. To this end, raw scores for verb fluency were defined as impaired if performance was at or below 1 standard deviation below the mean of the education matched published normative data for both actions (Piatt et al., 2004) and animal (Schretlen & Vannorsdall, 2010) fluency. Using these criteria, eight of the 20 participants had impaired action fluency performances, four had impaired animal fluency performances and three individuals had both impaired animal and action fluency performances.

For action fluency, percentage benefit from cueing was greater (Mean=58.87%) in those with impaired 60 second performances than in those without impairments (Mean = 9.19%; t(13.6)=−3.29, p<.01).

A similar finding was found in those who were impaired at on the 60 second animal fluency, where cueing helped those with impairments more than those without impairments (52.01% improvement in those with impaired animal fluency; 11.87% in the group without impaired animal performance (17)=−3.48 p=01).

Discussion

In the current study, we explored two questions related to action fluency impairments in pwPD. First, we evaluated whether apathy, in addition to motor dysfunction, may be related to action language difficulties in PD. Second, we explored the mechanism of fluency impairments in action and semantic fluency. We consider our findings in turn.

Apathy as a Substrate for Action Language Processing in PD

As we note in our introduction, embodied cognition theories have been supported by findings of verb processing deficits in those with motor disorders, such as pwPD (Salmazo-Silva et al, 2017; Barsalou, 2008; Barsalou, 2010; Damasio, 1989). However, current theories of language and semantic/conceptual processing (Martin 2016) suggest multiple substrates can underpin these functions. Persons with PD can have broader neurological effects as a result of their disease such as apathy, and potential relationships between such substrates and action language processing has been largely unexplored.

To this end, our study revealed moderate-sized and statistically significant relationships specifically between behavioral initiation aspects of apathy and cued verb generation. These findings were selective for action fluency and were not observed for semantic (animal) fluency performances. Combined with similar magnitude correlations between action fluency and motor measures, these results suggest that action language processing may be underpinned both by motor and apathy related substrates

There was the notable absence of a relationship between emotional/affective or cognitive/executive forms of apathy and verb generation tasks. This selectivity makes theoretical sense as behavioral/activation apathy is defined by reduced initiation of thoughts and behavior related to daily activities. This may represent a conceptually similar task to the action fluency task, where individuals are asked to self-generate actions; further work exploring action language as an objective marker of behavioral/activation apathy warrants further exploration

While tentative given the small sample size, this finding suggests that while PD has obvious and important motor dysfunction at its core, apathy may be another or shared source of verb grounding that contributes to dysfunction. This conclusion is strengthened by the lack of relationship between semantic fluency and motor performance or apathy. If this were just a general slowing effect or poor initiation in a subset of pwPD, then this same pattern would have been expected in semantic fluency data as well. Other ways to explore this possibility would be to compare structural connectivity metrics using neuroimaging in pwPD who either do or do not exhibit deficits and improvement with cueing on the verb fluency task. It would also be useful to determine whether action language is grounded in other sensory, motor, and emotional systems in order to give a comprehensive evaluation of the various ways language could be embodied in pwPD. Further, findings highlight the need to continue revisiting how verb processing relates to broader cognitive difficulties in pwPD, particularly executive and procedural memory processes (Ullman et al, 1997).

Retrieval, as opposed to Semantic Degradation Deficits, Appear to Underpin Verb Fluency Difficulties in PD

Our second aim was to explore the mechanism (lexical retrieval vs. semantic degradation) of action and animal fluency deficits in pwPD. By demonstrating improvements with cueing on action and animal fluency tasks, our results suggest lexical retrieval difficulties rooted likely in executive dysfunction (Galtier et al.,, 2017) rather than the wholesale loss of concepts or decreased access to lexical/conceptual information in pwPD. Indeed, pwPD who exhibited normatively defined clinical impairments in action or animal fluency improved around 50% with the provision of cues, a finding of much greater magnitude than their unimpaired peers.

This finding suggests the stimulability (i.e. ability to improve with cueing) of action language difficulties in pwPD and is an intriguing extension of embodied cognitive theories; making the body more salient may improve access to actions. In a related vein, it is notable that most individuals who were clinically impaired on semantic fluency also were clinically impaired on action fluency; however, a number of subjects showed impairments on action verb fluency only. Combined with a similar degree of improvement with cueing trials for animals and actions, these findings suggest that a shared retrieval deficit may underpin both action and animal fluency difficulties in PD, though the action impairments are indeed more pronounced and seemingly differentially related to motor dysfunction and apathy.

Limitations

This study has several notable limitations, one of which was the small and non-diverse sample. Testing these hypotheses in a broader sample, confirming the psychometric properties of the cued fluency measures created for this study, exploring more detailed motor assessments and their relationships to language measures, and having access to matched comparison groups will improve our understanding of these phenomena. Exploring the relationships observed herein with a non-PD matched control sample is particularly important for demonstrating whether findings are unique and specific to PD. Further, these results are correlational in nature, and broader samples, preferably with sufficient sample size to allow for multiple regression approaches to exploring these hypotheses and their interactions would be preferred. Such analyses would also allow for exploration of other cognitive variables that may covary with apathy, such as co-occurring executive dysfunction or broader memory retrieval difficulties which are common in pwPD. Finally, neuroimaging approaches, particularly neuroimaging connectivity metrics and task-based FMRI methods could be useful to to elucidate how variability in participants response patterns relate to core fronto-subcortical networks and to help more fully explore the relationship between apathy and action results.

Conclusions

Despite these limitations, the current study suggests that behavioral initiation apathy may be another contributor to the action language deficits in PD, and highlights how cueing may improve lexical access in PD.