A novel method for objective quantification of apathy based on gaze and physiological reactivity to stimuli presented in a virtual reality environment

Abstract

INTRODUCTION

We developed a tool for objective quantification of apathy.

METHODS

Participants (n = 97; 67 with cognitive impairment, 30 cognitively normal; mean age = 74.3 ± 6.2 years, 56.7% females) were exposed to emotional and cognitive stimuli in a virtual reality environment. Gaze metrics (time to first fixation [TTFF] and total fixation duration [TFD]) and autonomic nervous system (ANS) reactivity were measured. Apathy and depression were clinically assessed using the Lille Apathy Rating Scale short version and the Geriatric Depression Scale 15‐item version, respectively. Cutoffs of ≥ –7 and ≥ 5 were used to define apathy and depression, respectively.

RESULTS

The sample comprised 14 participants with apathy only, 9 with depression only, 10 with both, 63 with neither, and 1 with missing data. For all emotional stimuli, participants with apathy only showed longer TTFF (P = 0.039, effect sizes [ES] = 0.798), and shorter TFD (P = 0.023, ES = 0.578) compared to those without apathy or depression. ANS reactivity was not associated with apathy.

DISCUSSION

Apathy is associated with decreased gaze engagement at emotional stimuli.

Highlights

• Apathy measurement via questionnaires is limited by subjectivity biases.

• Apathy measurement via questionnaires is limited by simplistic scoring.

• We present a novel method for objective measurement of apathy.

• Gaze characteristics reflect the emotional and cognitive components of apathy.

1. INTRODUCTION

Apathy is a highly prevalent behavioral symptom in neurodegeneration, ¹ appearing in > 70% of patients with dementia, ² in > 40% of patients with mild cognitive impairment (MCI), ³ and in > 20% of non‐demented older adults. ⁴ It is a disorder of motivation, ¹ expressed as (1) narrowed emotional reactivity, (2) limited cognitive effort and interest, and (3) reduced goal‐directed activity ⁵ – reflecting its emotional, cognitive, and behavioral components, respectively. Apathy is associated with worse performance ⁶ and faster decline ⁷ in cognitive and functional abilities, greater impairments in brain structure and function, and with a higher burden of Alzheimer's disease and related disorders (ADRD) neuropathology. ⁸ , ⁹ , ¹⁰ Thus, apathy can potentially be used as a simple and accessible behavioral marker of incipient dementia and of a more aggressive disease course. Furthermore, by discouraging engagement in an active lifestyle—crucial for brain health ¹¹ —apathy itself is a therapeutic target.

The development of apathy as a marker of ADRD and as a treatment target ¹² is hindered by the limitation of the tools used for its quantification—primarily questionnaires administered to patients or their caregivers. These are prone to subjectivity biases induced by patients’ impaired self‐awareness, ¹³ and caregivers’ misinterpretations. ¹⁴ Additionally, questionnaires usually score apathy as a global entity, overlooking its distinct emotional, cognitive, and behavioral components. ¹⁵ Moreover, they do not necessarily facilitate the differentiation between apathy and depression, which symptomatically overlap ¹⁶ and frequently co‐occur. ¹⁶ Despite their potential contribution to apathy research, methods for objective apathy measurement are scarce and primarily capture (via movement sensors, video processing, measurement of facial expressions, ¹⁷ , ¹⁸ or speech analysis ¹⁹ ) motor or voice characteristics, ²⁰ which may also be affected by neurological symptoms other than apathy. ²¹ , ²² Moreover, these methods do not quantify the emotional and cognitive components of apathy.

Previous studies link gaze and autonomic nervous system (ANS) reactivity ²³ to emotional processing, ²⁴ prompting their use in the research of emotions ²⁵ in healthy ²⁶ and cognitively impaired populations. ²⁷ Gaze ²⁸ and ANS signals ²⁹ , ³⁰ , ³¹ have previously been correlated with apathy in patients with dementia, regardless of age and severity of cognitive impairment (CI). ²⁷ , ³⁰ , ³² , ³³ , ³⁴ Building on these connections, we created a tool to objectively measure apathy. Our aim was to examine the gaze and ANS correlates of clinically measured apathy (via questionnaire). In a virtual reality (VR) environment, individuals were exposed to emotional and cognitive stimuli while measuring their gaze and ANS (heart rate, galvanic skin response [GSR], and respiration) reactivity. ³⁵ VR was chosen because it elicits natural information processing, enabling ecologically valid experiments with high control. ³⁵

The scarce findings available today on objective correlates of clinically measured apathy have been more consistent on gaze characteristics ²⁷ , ²⁸ , ³² and less so on physiological measures. ²⁹ , ³⁰ , ³¹ , ³⁶ , ³⁷ Therefore, our primary hypothesis relates apathy status (clinically measured) to decreased gaze engagement at emotional stimuli. The hypotheses on physiological correlates of apathy (clinically measured) and cognitive statuses were exploratory.

2. METHODS

2.1. Participants

Participants (n = 97; 67 with cognitive impairment (CI) and 30 cognitively normal (CN)] were recruited from the Geriatric Psychiatry and Memory Unit atthe Sheba Medical Center, Israel. CN participants were usually spouses of CIparticipants or acquaintances of the study staff. Inclusion criteria: (1) Age 60‐90 (2) Diagnosis of mild‐moderate dementia or MCI ³⁸ for CI (3) Absence of significant emotional orbehavioral symptoms other than apathy or depression (assessed via theneuropsychiatric inventory ³⁹ ). Exclusion criteria: Severedementia (operationalized as a score ≤ 12 in the Mini Mental State Examinationscale ⁴⁰ ); (2) Clinical phenotypes that strongly suggest dementia subtypes other than AD (e.g., FTD, PSP, LBD) (3) Active medical problems potentially compromising compliance with the studyprotocol.

Ascertainment of MCI and mild–moderate dementia diagnosis was based on a full clinical evaluation, including collection of medical history, cognitive assessment, brain imaging, review of medical charts and lab testing as needed. The absence of cognitive impairment in control participants was based on a lack of dementia or MCI diagnosis in the medical records and on the cognitive assessment performed as part of the study protocol (see “see section 2.3.3”). The study was approved by the Sheba Medical Center Institutional Review Board (IRB).

2.2. Apparatuses

2.2.1. Virtual reality platform

Visual stimuli were presented in a fully immersive, first‐person perspective, VR naturalistic environment, using a head mount display (HMD) VR apparatus (HTC‐VIVE) with ≈ 100° and ≈ 110° field of view in the horizontal and vertical plane, respectively. After the participant had put on the HMD, a short unstructured interview was conducted to confirm that they were able to see the objects in the VR environment. In the VR environment (developed in Unity3D software), participants were sitting on a bench in a park, with one road in front of them. Thirty‐six emotional stimuli were presented to each participant (see section 2.3), either on billboards to the right and left side ahead of them (n = 18)—stationary areas of interest (AOI), or on bus ads (i.e., passing busses)—dynamic AOIs (n = 18). The stationary AOIs (20° wide and 16° tall) were presented for 10 seconds each. This timeframe was chosen to cover the range of physiological signals’ frequency ⁴¹ , ⁴² , ⁴³ while attempting to minimize the burden imposed on study participants in terms of the trial duration. To ensure the tool's capability to measure gaze reactivity, the billboards were situated 40⁰ to the left and right with respect to straight‐ahead sight lines, such that participants scan a large field of view. Henceforth, the bus stops were situated 55⁰ to the left and right (see Figure S1 in Supplementary Materials). The dynamic AOIs were introduced by moving from a distant point to a proximal bus stop within the virtual scene (this period lasted 30 seconds). Then, for an additional 8 seconds’ duration, the buses remained stationary at the bus stop (the dimensions of the bus ads during the stationary period were 10° to 15° wide and 15° to 30° tall). Finally, the bus proceeded in the opposite direction from which it had arrived (see demonstration video in Supplementary Materials). The time window that was used for eye tracking evaluations (see below), included 2 seconds prior to the stopping of the bus plus the 8 seconds while the bus was stationary at the bus stop.

2.2.2. Eye tracking system

To measure gaze reactivity, eye tracking was obtained within the HMD using an eye tracking device (Pupil Player, Pupil‐Labs) with a sampling rate of 200 Hz, a spatial accuracy of 1°, and a gaze precision of 0.08°. The eye‐tracking device records participants’ eyes with four sensors (two infrared cameras and two infrared illuminators). The systems provide the output of the gaze position coordinates as well as a video that overlays fixation points on the VR scenery. The plugin of the Pupil Player to Unity3D unifies between the coordinate system of the gaze with the virtual environment. In this process, the eye‐tracking data are down‐sampled to 50 Hz.

RESEARCH IN CONTEXT

1. Systematic Review: Literature review shows that apathy links to dementia risk and to brain pathology. Apathy measurement, primarily via questionnaires, is limited by subjectivity biases and by the global scoring of apathy, disregarding its distinct emotional, cognitive, and behavioral components. We developed a tool for objective quantification of the emotional and cognitive components of apathy, based on measurement of gaze and autonomic nervous system reactivity to stimuli presented in a virtual environment. We show that apathy is associated with lower gaze engagement at emotional and cognitive stimuli.

2. Interpretation: We present promising preliminary findings on an innovative and safe method for objective measurement of apathy.

3. Future Directions: Future studies should refine the virtual reality stimuli to optimize the quantification of subtle aspects of apathy components, while accounting for medical influences on the variables measured. A tool for objective quantification of the behavioral component of apathy should be developed.

2.2.3. Physiological signal measurement system

An EEGO‐Sport (ANT) device was used to measure ANS reactivity. This is a portable wearable wireless system that collects physiological bio signals at a sampling rate of 512 Hz, covering the frequency ranges of the functions measured, that is, 0.5 to 40 Hz, ⁴¹ 0.05 to 5 Hz, ⁴³ and 0.15 to 0.6 Hz ⁴² for echocardiogram (ECG), GSR, and respiration, respectively.

2.3. Procedure

Objective (via VR) and subjective (via the Lille Apathy Rating Scale [LARS], ⁴⁴ short version ⁴⁵ ) assessments of apathy were conducted on the same day in the vast majority of participants except for n = 6, in which the time that elapsed between assessment varied between 2 and 17 days.

To objectively evaluate the emotional component of apathy, three types of emotional stimuli (12 per type) were randomly presented: (1) positive (pictures of participants’ grandchildren), (2) aversive (bleeding organ, a human body in a motor vehicle accident scene), and (3) neutral (a chair). The two latter types of stimuli were sourced from the International Affective Picture System inventory. ⁴⁶ Each picture was displayed for 10 seconds, with a 30‐second interval between stimuli to allow the return of ANS signals to baseline. Participants’ ANS and gaze reactivity to these stimuli were measured (see section 2.4).

To objectively assess the cognitive component of apathy, cognitive stimuli were presented on stationary billboards, situated 30⁰ to the left and right with respect to a straight‐ahead sight line (see Figure S2 in Supplemetary Materials). Each stimulus consisted of two simultaneously presented pictures, one congruent (e.g., a chicken) and one incongruent (e.g., a chicken with a human head), ⁴⁷ allowing examination of gaze preference. Picture pairs (n = 20) were presented for 10 seconds each, with a 10‐second break between stimuli.

2.4. Outcome measures

All data analysis and preprocessing were performed with MATLAB (The MathWorks Inc.).

2.4.1. Gaze reactivity (obtained from eye‐tracking analyses)

We measured time to first fixation (TTFF)—the time that elapsed between stimulus presentation until gaze fixation at it, and total fixation duration (TFD)—the duration of gaze fixation on the stimulus.

To quantify the cognitive component of apathy, for each pair of cognitive stimuli, we calculated the relative time spent gazing on the incongruent stimulus (RTSI) using the following percentile parameter:

$$\mathrm{RTSI}\left(\%\right)=100\ast \frac{Tinco}{Tinco+Tco}$$ (1)

where Tinco and Tco are the durations of gaze fixation on the incongruent and congruent pictures, respectively. Next, for each participant, we calculated the mean RTSI for all stimuli pairs.

2.4.2. ANS reactivity

Measurement was based on the differences between the average value of each physiological signal during the 10 seconds prior to and the average of the values during the 10 seconds after gaze fixation. Each of these measures was normalized to the corresponding value during 10 seconds prior to the first gazing time (“baseline”) according to the following formula:

$$\mathrm{Measure}\left(\%\right)=100\ast \frac{\mathrm{Value}\mathrm{after}\mathrm{stimulus}-\mathrm{Baseline}\mathrm{value}}{\mathrm{Baseline}\mathrm{value}}$$ (2)

Reactivity of the following signals was collected: GSR, ECG, and respiration.

Preprocessing of GSR data was performed using “Ledalab” plugin. ⁴⁸

Heartbeat peak timings were extracted from the raw ECG signal using the “Pan Tompkins” algorithm. ⁴⁹ To characterize heartbeat fluctuations, we calculated heart rate variability (HRV) measures in the time domain, namely the standard deviation of consecutive inter‐beat intervals (HRV_SDNN), reflecting mainly parasympathetic activity, and the root mean square of successive inter‐beat interval differences (HRV_RMSSD), reflecting sympathetic and parasympathetic activities. ⁵⁰

Respiratory peak timings were extracted from the raw respiration signal after applying a lowpass filter of 4 Hz. We calculated the respiration rate and the respiration depth to characterize changes in respiratory cycles. Details on missing data can be found in the Supplementary Materials.

2.4.3. Clinical assessments

1. Cognitive screening was carried out via the Montreal Cognitive Assessment (MoCA), ⁵¹ Hebrew version. ⁵²

2. Apathy was subjectively measured via the LARS ⁴⁴ short version, ⁴⁵ based on an interview with the study participant. It comprises 12 queries in seven domains, each corresponding to one of the clinical manifestations of apathy: everyday productivity, interests, taking initiative, novelty seeking, motivations, voluntary actions, emotional responses, and social life. LARS scores’, short version, range between −15 (absence of apathy) to +15 (highest degree of apathy). A cutoff of ≥ −7 defines presence of clinically significant apathy.

3. Depression was subjectively measured via the Geriatric Depression Scale (GDS), 15‐item version, ⁵³ a self‐reported scale, with scores ranging from 0 to 15, higher scores indicating more depression. A cutoff of ≥ 5 was used to define clinically significant depression.

4. Additionally, participants rated their VR experience via the short feedback questionnaire (SFQ), ⁵⁴ a scale ranging between 0 and 4 (“0” = not immersive; “4” = fully immersive), their enjoyment during the experiment, and the difficulty on a scale ranging between 0 and 4 (“4” = maximal value).

2.5. Statistical analyses

Statistical analysis was performed using SPSS 2020 Statistical software (IBM). The significance level was set to α = 0.05.

2.6. Hypotheses

2.6.1. Primary hypothesis

For all stimuli, TTFF will be longer and TFD will be shorter in participants with apathy compared to participants without apathy or depression.

2.6.2. Secondary hypothesis

For all stimuli, TTFF will be longer and TFD will be shorter for CI patients compared to CN. Additionally, in exploratory analyses, we examined whether the presence of apathy and CI are associated with physiological reactivity, as reflected by changes in ANS signals, in response to stimuli presented in the VR environment. We hypothesized that for all emotional stimuli, physiological reactivity will be reduced (1) in participants with apathy compared to participants without apathy or depression; (2) in CI participants compared to CN.

To examine the contribution of cognitive status to the relationship of clinically measured apathy with gaze and physiological reactivity characteristics, we examined these relationships in the group of participants with CI. We hypothesized that for all stimuli, TTFF will be longer, TFD will be shorter, and physiological reactivity will be reduced in participants with CI with apathy compared to CI patients without apathy or depression.

We also examined whether gaze and physiological reactivity to different types of stimuli differ between apathy and depression. Based on previous studies showing that depression is associated with difficulty in disengaging from negative stimuli, ⁶ we hypothesized that gaze engagement and physiological reactivity to aversive stimuli would be increased in participants with compared to those without depression.

For all hypotheses, we separated the models for gazing behavior parameters from the ANS outcome measures. We applied a separate model for each comparison including only patients from the two groups of interest in each hypothesis. The models used were as follows: for all endpoints except for TTFF we used a linear mixed effect regression model including the independent variables: group indicator (apathy or cognitive status), condition type (type of stimulus), TFD (except when TFD is the outcome), and subject as a random effect. The groups (group indicator) used in the different analyses, their sizes, and the type of stimuli that were used for measurement of physiological response are described in Table S1 in supplementary materials. We applied the log transformation to obtain a more normal distribution (see Figures S3–S5 in supplementary materials). TTFF is a time‐to‐event variable, censored at 10 seconds. Cox regression was used to analyze these data, as it accommodates censored observations and is appropriate for time‐to‐event analysis. ⁵⁵ , ⁵⁶ The model included the same independent variables as previous analyses, excluding TFD. The subject's random effect was incorporated using a frailty model. We controlled the Type 1 error for multiple testing only for the two primary outcomes, TFD and TTFF, within the primary and secondary analyses separately. All other tests were regarded as exploratory and no correction for multiple comparisons was done. Effect sizes (ES) of type d were calculated as the statistic t of the group comparison divided by 2 times the square root of its degrees of freedom.

Last, for the cognitive component of apathy, we posited that participants with apathy will devote an equal amount of gazing time at congruent and incongruent pictures whereas those without apathy will devote a larger amount of time to gaze at incongruent stimuli.

We expected similar trends in CI compared to CN participants. We applied a repeated measures analysis of variance model for group effect (apathy only vs. no apathy or depression, regardless of cognitive status and CI vs. CN).

3. RESULTS

Compared to CN, CI participants had lower MoCA scores (P < 0.001), lower education (P = 0.014), higher GDS scores (P = 0.033), and higher LARS scores (P = 0.001; Table 1).

TABLE 1.

Participants’ characteristics (mean ± SD) by cognitive status.

	CN	CI	pvalue
N	30	67 a
Age (years)	73.7 ± 5.31	74.61 ± 6.51	0.647
Sex (% female)	66.66%	52.24%	0.185
Education (years)	15.93 ± 1.98	14.48 ± 3.09	0.014 *
MoCA score	26 ± 2.51	19.62 ± 5.37	<0.001 *
LARS score	−11.96 ± 2.93	−9.38 ± 3.90	0.001 *
GDS score	1.8 ± 2.01	2.92 ± 2.67	0.033 *

Abbreviations: CI, cognitively impaired; CN, cognitively normal; GDS, Geriatric Depression Scale; LARS, Lille Apathy Rating Scale; MoCA, Montreal Cognitive Assessment; SD, standard deviation.

^a Physiological signals from one CI participant were technically not sound and thus not included in the analyses. Data available for the cognitive scene are from 19 CN to 47 CI.

^* P ≤ 0.05, Mann–Whitney U test. Ratios were compared using the chi‐square test.

Of the entire sample, n = 24 (24.74%) and n = 19 (19.58%) reached LARS and GDS cutoffs for apathy and depression, respectively, and n = 10 (10.31%) exhibited above‐threshold scores on both scales (Table 2). The prevalence of both apathy and depression was higher among participants with CI (Table 2). GDS and LARS scores were positively correlated (r _s = 0.294, P = 0.004). Irrespective of cognitive status, participants with apathy (LARS score ≥ −7) but without depression (i.e., GDS < 5) did not differ from participants without apathy and without depression (Table 3) in demographic and clinical characteristics other than apathy score. For individual participants’ characteristics, please see Table S2 in supporting information.

TABLE 2.

Number (and %) of participants with above threshold scores for apathy and depression by group.

	Total sample		CN		CI
	GDS ≥ 5 n (%)	LARS ≥ −7 n (%)	GDS ≥ 5 n (%)	LARS ≥ −7 n (%)	GDS ≥ 5 n (%)	LARS ≥ −7 n (%)
GDS ≥ 5 n (%)	19 (19.58)	10 (10.31)	3 (10.00)	2 (6.66)	16 (23.88)	8 (11.94)
LARS ≥ −7 n (%)	10 (10.31)	24 (24.74)	2 (6.66)	4 (13.30)	8 (11.94)	20 (29.85)

Abbreviations: CI, cognitively impaired; CN, cognitively normal; GDS, Geriatric Depression Scale; LARS, Lille Apathy Rating Scale.

TABLE 3.

Participants’ characteristics (mean ± SD) by apathy status. ^a

	AP−	AP+	AP+ and depression	P value
N	63 a	14	10
Age (years)	74.41 ± 5.64	72.57 ± 6.11	74.7 ± 8.69	0.233
Sex (% female)	57.14%	28.57%	80%	0.053
Education (years)	15.75 ± 2.58	14.31 ± 2.32	13.1 ± 3.21	0.101
LARS score	−12.11 ± 2.21	−4.78 ± 2.15	−5.2 ± 3.01	<0.001 *
GDS score	1.57 ± 1.29	1.42 ± 1.28	6.9 ± 1.97	0.726
MoCA score	22.75 ± 4.89	20 ± 6.93	19.6 ± 6.55	0.199

Abbreviations: AP+, apathy only; AP−, no apathy or depression; AP+ and depression, apathy and depression; GDS, Geriatric Depression Scale; LARS, Lille Apathy Rating Scale; MoCA, Montreal Cognitive Assessment.

^a Irrespective of cognitive status; all without depression. Physiological signals from one CI participant were technically not sound and thus not included in the analyses. Data available for the cognitive scene (see below) are from 44 AP− and 10 AP+.

^* P ≤ 0.05, Mann–Whitney U test. Ratios were compared using the chi‐square test.

3.1. VR paradigm feasibility

The VR procedure was generally well tolerated and accepted by the participants (see Supplementary Materials on acceptability of VR procedure by participants).

3.2. Gaze and ANS reactivity to emotional stimuli

3.2.1. The association of apathy status with gazing behavior: primary hypothesis

For all types of stimuli, TTFF was longer (χ ² = 4.25; df = 1; P = 0.039; ES = 0.798) and TFD was shorter (t = 2.33; df = 65; P = 0.023, ES = 0.578) in participants with apathy (N = 14) compared to participants without apathy or depression (N = 62; mean TTFF 1.52 ± 2.00 vs. 1.21 ± 1.73 and mean TFD 5.94 ± 2.89 vs. 6.73 ± 2.77 seconds in participants with and without apathy, respectively; Figure 1 and Tables S3 and S4 in supplementary materials).

FIGURE 1.

Gaze behavior by apathy status: (A) TFD and (B) TTFF. In each graph, data are plotted separately for each group (red and blue colors represent participants with apathy only and participants without apathy or depression, respectively), with boxplots depicting the overall group median and lower and upper hinges corresponding to the first and third quartiles (the 25th and 75th percentiles). Maximal and minimal values are represented by horizontal cups. AP+, apathy only; AP−, no apathy or depression; TFD, total fixation time; TTFF, time to first fixation

3.2.2. The association of cognitive status with gazing behavior: secondary hypothesis

For all types of stimuli, TTFF was longer (χ ² = 10.72; df = 1; P = 0.001; ES = 0.777) and TFD was shorter (t = 2.60; df = 83; P = 0.011; ES = 0.572) in CI participants (N = 66) compared to CN participants (N = 30; mean TTFF 1.43 ± 1.95 vs. 0.99 ± 1.49 and mean TFD 6.27 ± 2.89 vs. 6.92 ± 2.71 seconds in CI and CN respectively; Figure 2, Tables S3 and S4).

FIGURE 2.

Gaze behavior by cognitive status: (A) TFD and (B) TTFF. In each graph, data are plotted separately for each group (red and blue colors represent participants with cognitive impairment and participants with normal cognition, respectively), with boxplots depicting the overall group median and lower and upper hinges corresponding to the first and third quartiles (the 25th and 75th percentiles). Maximal and minimal values are represented by horizontal cups. CI, cognitively impaired; CN, cognitively normal; TFD, total fixation duration; TTFF, time to first fixation

3.2.3. The association of apathy and cognitive statuses with ANS and gaze reactivity (exploratory)

ANS reactivity measures did not differ when comparing participants with apathy to those without apathy or depression, or when comparing CN to CI participants (P ≥ 0.121 for all ANS measures; see Tables S3–S6 in supplementary).

We then examined gaze and ANS reactivity to specific types of emotional stimuli. Irrespective of cognitive status, (1) in participants with apathy and without depression (n = 14), TFD was longer for aversive stimuli compared to participants with depression but without apathy (n = 9); (2) in participants without depression or apathy (n = 62), TFD was longer for aversive stimuli compared to participants with depression only (t = −2.56; df = 19; P = 0.019, ES = −1.174 and t = 2.89; df = 60; P = 0.005; ES = 0.746, respectively; Table S6). Similarly, when examining these relationships only within the group of CI participants, (3) in those with apathy but no depression (n = 12), TFD was longer for aversive stimuli compared to those with depression but no apathy (n = 8; t = −2.89; df = 16; P = 0.011, ES = −1.446; Table S6). No statistically significant effects were found for all other outcome measures, or for all other subgroup comparisons (P ≥ 0.094; Table S6).

3.2.4. The associations of apathy and depression with gaze characteristics, treating LARS and GDS scores as continuous variables (exploratory)

For the entire cohort, TFD at positive images was negatively correlated with LARS scores (r _s = −0.234, P = 0.032; Figure 3A) and TFD at negative images was negatively correlated with GDS scores (r _s = −0.267, P = 0.013; Figure 3B). Other statistically significant correlations between gaze and LARS or GDS scores were not found.

FIGURE 3.

The association of TFD with apathy and depression scores. (A) TFD at positive images versus LARS score. (B) TFD at negative images versus GDS score. In each graph, data are plotted with a scatter plot depicting individual mean values. Each group is represented by different colors (red and blue dots represent participants with cognitive impairment and participants with normal cognition, respectively). CI, cognitively impaired; CN, cognitively normal; GDS, Geriatric Depression Scale; LARS, Lille Apathy Rating Scale; TFD, total fixation duration.

Within participants with CI, we observed similar trends for the correlations between TFD at positive images and LARS scores (r _s = −0.149, P > 0.26) and for the correlation between TFD at negative images and GDS score (r _s = −0.234, P > 0.071). However, these did not reach statistical significance.

3.3. Gaze reactivity to cognitive stimuli

3.3.1. The association of apathy status with gaze characteristics

We compared data from participants with apathy (N = 10) to those without (N = 44) apathy or depression, regardless of cognitive status. RTSI did not differ between groups (F _1,52 = 2.742; P = 0.104; ηp² = 0.05). When LARS scores were treated as a continuous variable, higher LARS scores were associated with shorter TFD at both congruent (r = −0.244; P = 0.031) and incongruent (r = −0.224; P = 0.048) pictures.

4. DISCUSSION

We propose a novel method for objective quantification of apathy using gaze and ANS reactivity to stimuli presented in a VR environment. Our findings support our primary and secondary hypotheses, demonstrating that for all types of emotional stimuli combined, the presence of apathy, irrespective of cognitive status, as well as CI per se, are associated with prolonged TTFF and shortened TFD, suggesting reduced gaze engagement at emotional stimuli. Further, exploratory analyses revealed the potential of this tool to differentiate between apathy and depression such that gazing time at aversive stimuli was shorter in participants with depression compared to participants with apathy or those with neither apathy nor depression. Additionally, when examining LARS and GDS scores as continuous variables, higher TFD at positive images correlated with lower LARS scores, while higher TFD at negative images correlated with lower GDS scores.

Our study confirms a link between clinically measured apathy and shorter gaze fixation at emotional stimuli, ²⁷ supporting the potential of this method for objective quantification of the emotional component of apathy. As for the cognitive component of apathy, we found an association between clinically measured apathy and shorter gaze fixation at both congruent and incongruent emotional stimuli. This is in contrast to previous studies ⁵⁷ in which apathy was associated with reduced gaze engagement at incongruent cognitive stimuli and even distribution of gazing time between congruent and incongruent stimuli. Although methodological differences might explain these discrepancies, the findings highlight the need for further research on the nuances of apathy and its distinct components. Additionally, the factors driving our observations (e.g., difficulty in set shifting or in retaining attention, changes in brain circuits responsible for emotional response) should be studied.

In contrast to some, ²⁹ , ³⁰ but not all, ³³ previous studies, ANS reactivity to emotional stimuli was not associated with clinically measured apathy. In our paradigm, emotional stimulation was based on pictures rather than movies. Although several types of stimuli were presented (positive, neutral, and aversive), these did not encompass the entire spectrum of positive and negative emotions. ⁵⁸ The timeframe used for measurement of physiological measures may have been suboptimal for some physiological signals, thus restricting our sensitivity. Finally, we cannot rule out the effect of medical factors on ANS reactivity. Future studies will investigate if questionnaire‐based apathy measures correlate with ANS activity and represent the same clinical construct.

Previous studies linked negative emotions to increased focus on negative stimuli. ⁶ In the present study, depression was associated with shorter, not longer, gaze on negative images. These conflicting findings may suggest that different negative emotions (e.g., depression vs. anxiety) might impact attention bias differently. ⁵⁹

The method presently proposed for apathy measurement offers several key advantages. We used a fully immersive VR environment, shown to elicit more natural gaze behavior compared to traditional computer screens, ³⁵ thus facilitating ecologically valid evaluations while maintaining control over variables. The method can potentially overcome the subjectivity biases inherent to apathy measurement questionnaires. ⁶⁰ Additionally, the emotional and cognitive components of apathy, each linked to distinct frontal lobe circuits, ¹⁵ are quantified separately, facilitating future exploration of their distinct clinical and prognostic significances. Finally, the procedure is safe and tolerated, making it scalable for wider use.

This study has several limitations. The small sample size, low apathy scores, and overlapping gaze patterns between groups with and without apathy may have limited statistical power. However, our findings suggest the tool's potential to detect subtle apathy levels. Future studies with larger, clinically apathetic samples will be needed to confirm these results and assess the physiological markers' discriminant validity. Definition of CN status was based on medical records and MoCA scores rather than a full clinical evaluation. Apathy and cognitive statuses were similarly associated with gaze characteristics, hindering the isolation of apathy's specific effect. However, the relationship between TFD at aversive stimuli and apathy in the entire sample paralleled that relationship observed in the CI group, suggesting the potential specificity of this tool for identifying apathy per se. Nevertheless, the role of cognitive and sociodemographic factors in the observed relationships requires further investigation. The VR environment and physiological measurement system may have caused discomfort or distraction, potentially affecting results.

Optimizing stimulus presentation (duration and type of stimuli), and physiological signals’ measurement methods while considering dementia status and etiology, mixed states of apathy and depression, medical comorbidities, medications, and visual acuity will be crucial in future studies. Clinical evaluations of apathy and depression should be expanded in future research to account for informant reports and diagnostic criteria.

A platform for objective quantification of the behavioral and social components of apathy is presently not available.

In conclusion, we present promising preliminary findings on an innovative and safe method for objective measurement of subtle apathy forms. The scalability of this method aligns with the anticipated increasing simplicity and accessibility of VR platforms, positioning it as a promising tool for clinical and research applications.

CONFLICT OF INTEREST STATEMENT

The authors have nothing to disclose. Author disclosures are present in supporting information.

CONSENT STATEMENT

All human subjects provided informed consent.

Supporting information

Supporting Information

Supporting Information

Ravona‐Springer R, Koren O, Galor N, et al. A novel method for objective quantification of apathy based on gaze and physiological reactivity to stimuli presented in a virtual reality environment. Alzheimer's Dement. 2025;17:e70020. 10.1002/dad2.70020