INTRODUCTION
Neurodegenerative diseases (NDDs) such as Alzheimer’s disease (AD) or Huntington’s disease (HD) inevitably lead to impairments in brain functions that are crucial for daily life functioning. For instance, is has been shown that a decline in cognitive functions such as memory, attention, and executive function can have severe consequences for the abilities required for critical judgment and adequate decision-making, which are important in risky or hazardous traffic situations (Alameda-Bailén, Salguero-Alcañiz, Merchán-Clavellino, & Paíno-Quesada, Reference Alameda-Bailén, Salguero-Alcañiz, Merchán-Clavellino and Paíno-Quesada2017; Cahn-Weiner, Boyle, & Malloy, Reference Cahn-Weiner, Boyle and Malloy2002).
Increasingly, research is focusing on deficits in social cognition in various neurodegenerative disorders and their impact on everyday life (Christidi, Migliaccio, Santamaría-García, Santangelo, & Trojsi, Reference Christidi, Migliaccio, Santamaría-García, Santangelo and Trojsi2018). Social cognition involves the brain functions which allow us to behave adequately in social situations. It is a broad construct comprising different aspects (Adolphs, Reference Adolphs2009). These different aspects include the ability to perceive socially important information (such as the perception of emotional facial expressions), the ability to understand intentions and behavior of others (such as creating a theory of mind) and empathic behavior. Neural substrates of social cognition have been found to comprise frontal-subcortical circuits, including the orbitofrontal (Tekin & Cummings, Reference Tekin and Cummings2002) and ventromedial prefrontal cortex (Shamay-Tsoory, Tomer, Berger, & Aharon-Peretz, Reference Shamay-Tsoory, Tomer, Berger and Aharon-Peretz2003). Furthermore, medial temporal lobe structures, including the superior temporal sulcus and amygdala, have been identified as important brain areas related to social cognition (Carrington & Bailey, Reference Carrington and Bailey2009). NDDs often lead to damage to these frontal-subcortical and temporal brain areas, both at a cortical level, such as AD (Tekin & Cummings, Reference Tekin and Cummings2002) and frontotemporal dementia (FTD; Rosen et al., Reference Rosen, Gorno-Tempini, Goldman, Perry, Schuff, Weiner and Miller2002), and at a subcortical level, such as HD (Liu, Yang, Burgunder, Cheng, & Shang, Reference Liu, Yang, Burgunder, Cheng and Shang2016). Accordingly, impairments in social cognition have been found in patients with various NDDs, including HD, AD, FTD, and Dementia with Lewy Bodies (DLB; Desmarais, Lanctôt, Masellis, Black, & Herrmann, Reference Desmarais, Lanctôt, Masellis, Black and Herrmann2018; Henley et al., Reference Henley, Novak, Frost, King, Tabrizi and Warren2012; Kumfor et al., Reference Kumfor, Irish, Leyton, Miller, Lah, Devenney and Piguet2014).
One of the most important aspects of social cognition is the recognition of emotional facial expressions. It is has been found that the recognition of emotions and the experience of emotions are closely related (Enticott, Johnston, Herring, Hoy, & Fitzgerald, Reference Enticott, Johnston, Herring, Hoy and Fitzgerald2008). Hence, impairments in the ability to recognize facial expressions may be a marker of a decreased ability to experience emotions. A decreased ability to experience emotions can, in turn, interfere with adequate decision-making behavior (Damasio, Tranel, & Damasio, Reference Damasio, Tranel and Damasio1991). In particular, the experience of fear has been found to guide decision behavior in risky situations (Öhman, Reference Öhman2005). Consequently, it seems plausible that impaired recognition of fear, as a probable marker of an impaired experience of fear, is related to increased risky decision behavior. This was indeed found in a study with patients with traumatic brain injury (TBI; Visser-Keizer, Westerhof-Evers, Gerritsen, Van der Naalt, & Spikman, Reference Visser-Keizer, Westerhof-Evers, Gerritsen, Van der Naalt and Spikman2016). TBI is a neurological disorder which often includes damage to the prefrontal brain circuits underlying emotion recognition (Fujiwara, Schwartz, Gao, Black, & Levine, Reference Fujiwara, Schwartz, Gao, Black and Levine2008). Visser-Keizer et al. (Reference Visser-Keizer, Westerhof-Evers, Gerritsen, Van der Naalt and Spikman2016) found a significant correlation between worse recognition of fearful facial expressions and risky decision-making, using the Iowa Gambling Task (IGT), an experimental task, as a measure of risk-taking behavior (Bechara, Damasio, Damasio, & Anderson, Reference Bechara, Damasio, Damasio and Anderson1994). In addition, in patients with cerebellar stroke, a relationship between fear-recognition and risk-taking behavior in traffic situations was found (van den Berg, Huitema, Spikman, Luijckx, & de Haan, Reference van den Berg, Huitema, Spikman, Luijckx and de Haan2020).
To date, only one study investigated emotion recognition and risky decision-making behavior in patients with an NDD, that is, Parkinson’s disease, and indeed found a significant relationship between worse emotion recognition and performance on the IGT (Ibarretxe-Bilbao et al., Reference Ibarretxe-Bilbao, Junque, Tolosa, Marti, Valldeoriola, Bargallo and Zarei2009). However, there is a lack of studies investigating the role of impaired emotion recognition, in particular of fearful expressions, in risk-taking behavior in more natural, everyday life tasks, such as traffic situations, in patients with NDDs. Therefore, the aim of the present study was to assess the role of emotion recognition, in particular of fear, in risky decision-making in traffic-related tasks, in a diverse group of patients with NDDs, including AD, FTD, DLB, and HD, for whom it is well known that those brain circuits that are relevant for emotion perception are or will be affected during the course of the disease (Elamin, Pender, Hardiman, & Abrahams, Reference Elamin, Pender, Hardiman and Abrahams2012). Furthermore, we sought to evaluate under which circumstances this relationship could exist, such as differences between traffic situations with and without a direct danger. Investigating factors related to unsafe decision-making is crucial for a timely and valid assessment of fitness to drive of patients with (early stages of) NDDs.
METHODS
Design and Setting
This study was a retrospective review of records, in which data were combined from patients and healthy controls (HCs) who took part in two larger studies investigating AD, FTD, DLB, and Mild Cognitive Impairment (MCI; ‘DUAL-Positron Emission Tomography (PET) in Dementia’) and HD (‘Fitness to Drive in Huntington’s Disease’). Both study protocols were approved by the Medical Ethical Committee of the University Medical Center (number 2014.320 and 2015.310, respectively). The patients of the DUAL-PET study were recruited through a collaboration of ten memory clinics (The North Netherlands Memory Research Network). Patients in local centers were referred to the Alzheimer Research Center of the University Medical Center Groningen (UMCG). HD patients were approached by their clinician for inclusion. All participants were seen at the UMCG, Department of Neurology, Unit Neuropsychology.
Furthermore, HCs were included in order to compare the test results between patients and HCs. HCs were volunteers, recruited by advertisements in the Alzheimer Research Center and by online advertisements. They were also tested in the UMCG. HCs performed the same test protocol as the patient group. In addition to this, extra HC data from previous studies at the Department of Neurology were available for the emotion recognition task and the risk-taking behavior task. These additional HCs had participated in related studies at the Department of Neurology, which involved a neuropsychological assessment of social cognition tests of approximately 1 to 1.5 h.
All patients and HCs who took part in driving-related studies had to have a car driver’s license in order to be eligible for inclusion for this study. Participants had to be “active drivers”: they should have driven at least 300 km in the last year. This criterion has also been used in previous studies (e.g., Jacobs et al., Reference Jacobs, Hart, Mejia Miranda, Groeneveld, van Gerven and Roos2019). Exclusion criteria for all participants consisted of substance abuse, serious psychiatric, or neurological disorders (other than an NDD in case of patients) or insufficient command of the Dutch language. All participants signed a written informed consent and were treated in accordance with the Declaration of Helsinki.
Participants
Fifty-one patients diagnosed with an NDD were included. The group of patients was diagnosed according to the current clinical criteria and consisted of 31 patients diagnosed with AD (19 with AD dementia (McKhann et al., Reference McKhann, Knopman, Chertkow, Hyman, Jack, Kawas and Phelps2011) and 12 with MCI due to AD (Albert et al., Reference Albert, DeKosky, Dickson, Dubois, Feldman, Fox and Phelps2011)), 4 with FTD (Rascovsky et al., Reference Rascovsky, Hodges, Knopman, Mendez, Kramer, Neuhaus and Miller2011), 7 with DLB (McKeith et al., Reference McKeith, Dickson, Lowe, Emre, O’Brien and Feldman2005), and 9 patients with HD (8 with symptomatic HD and 1 with pre-symptomatic HD) (Kieburtz, Reference Kieburtz1996). Diagnoses of AD, FTD, and DLB were, apart from neuropsychological assessment, confirmed with structural magnetic resonance imaging, cerebrospinal fluid biomarkers, Pittsburgh compound B PET, and 2-[18 F]fluoro-2-deoxy-D-glucose PET. Diagnoses of HD were confirmed by genetic analysis.
Measures
A neuropsychological assessment was performed to assess emotion recognition and risk-taking behavior. The following tests were performed in this order: The Ekman 60 Faces Test of the Facial Expressions of Emotions – Stimuli and Test (FEEST), The Action Selection Test (AST), and three drives on a driving simulator. Furthermore, a subgroup of patients performed a few additional neuropsychological tests, such as the Trail Making Test, a measure of psychomotor speed (Reitan, Reference Reitan1958). The total testing time took approximately 1–1.5 h.
Emotion recognition
The Ekman 60 Faces Test of the FEEST (Young, Perrett, Calder, Sprengelmeyer, & Ekman, Reference Young, Perrett, Calder, Sprengelmeyer and Ekman2002) is a sensitive and commonly used measure of emotion recognition (Voncken, Timmerman, Spikman, & Huitema, Reference Voncken, Timmerman, Spikman and Huitema2018). Sixty faces displaying one of the six basic emotions (anger, disgust, fear, happiness, sadness, and surprise) are shown, one at a time. Every emotion expression is presented 10 times for 3 s in a randomized order, after which six labels are presented and the participant is asked to indicate which emotional expression was shown. The six basic emotion scores (FEEST-Ang, FEEST-Disg, FEEST-Fear, FEEST-Hap, FEEST-Sad, and FEEST-Sur) range from 0 to 10, and the total score (FEEST-Tot) ranges from 0 to 60. Higher scores on the FEEST indicate better performance.
The Action Selection Test
The AST from the SWOV Institute for Road Safety Research (Vlakveld, Reference Vlakveld2011) was used to measure risk assessment and hazard anticipation in traffic situations, which has also been used by Van den Berg et al. (Reference van den Berg, Huitema, Spikman, Luijckx and de Haan2020) and by Piersma et al. (Reference Piersma, Fuermaier, de Waard, Davidse, de Groot, Doumen and Tucha2016). Note that this same test has been referred to as the “hazard perception test” by Piersma et al. (Reference Piersma, Fuermaier, de Waard, Davidse, de Groot, Doumen and Tucha2016). In this test, 25 pictures of different traffic situations from a car driver’s perspective are presented on a computer screen. The situations contain direct dangers, such as a child running across the street, which have to be directly evaluated and anticipated. Participants are asked to indicate what they would do in that particular situation: do nothing, release accelerator, or brake. Every picture is shown for 8 s, but participants could respond after these 8 s. This long exposure time was deliberately chosen by Vlakveld et al. (Reference Vlakveld2011), as the test was developed with the aim of assessing risk-taking behavior in situations where it would be expected that the hazard was detected. In contrast to real driving behavior, the situation is completely new to the participants in a static picture; the static pictures do not show how situations have developed. Hence, participants may have needed more time to detect the potential hazard in the static pictures of the AST. The responses were rated by experts as very safe, safe, correct, unsafe, or very unsafe (Vlakveld, Reference Vlakveld2011). We assigned zero points to correct or (very) safe responses, one point to unsafe responses, and two points to very unsafe responses. The outcome measure was the sum score of unsafe and very unsafe responses, where a higher score indicates more risk-taking behavior. This score, termed AST, was used as a measure of risk behavior in situations involving anticipating a direct danger.
Driving simulator tasks
Two scenarios in an advanced driving simulator were used to measure risky decision-making behavior: (1) the swing drive, a scenario without the presence of direct dangerous situations and (2) the intersections drive, a scenario involving anticipating direct dangerous situations.
The driving simulator used for the current study has also been used in previous studies (e.g., Dotzauer, Caljouw, De Waard, & Brouwer, Reference Dotzauer, Caljouw, De Waard and Brouwer2013; Dotzauer, Caljouw, De Waard, & Brouwer, Reference Dotzauer, Caljouw, De Waard and Brouwer2015; Dotzauer, De Waard, Caljouw, Pöhler, & Brouwer, Reference Dotzauer, De Waard, Caljouw, Pöhler and Brouwer2015). The configuration of the driving simulator consists of a projection screen stage and an open cabin “mock-up” standing within this stage. It contains a custom-made force-feedback steering wheel, accelerator, clutch, and brake pedals. The projection screen stage consists of three video projectors and a large flat-screen of 6-meter length, bent 60 degrees inwards at 1/3 and 2/3 of its length, which presents the driver with a wide horizontal view. The software used in the simulator was from St Software. Both drives require driving on a winding two-lane country road with alternating left–right curves of 40 degrees and 500-meter radius, while there is a continuous stream of traffic from the opposing direction, but no traffic on the lane of the simulator drives. The car was set to automatic transmission.
Prior to the swing drive and the intersections drive, participants performed an exercise drive, the fixed drive, to get used to the simulator. In the fixed drive, which takes approximately 5 min, participants are instructed to drive safely in the middle of their lane. The speed of the car is automatically regulated (“fixed”) in this drive, so participants only have to steer during this drive, to get used to simulator environment and the steering wheel. In particular, they will be able to experience the sensitivity of the steering wheel in order to know how to stay in the middle of their lane. The steering wheel vibrates when participants get off the road. The speed is gradually increased during this drive, so that participants can get a sense of how fast a particular speed would feel like. The speed starts at 50 kilometers per hr (km/hr) and increases in steps of 10 km/hr until 100 km/hr.
In the swing drive, which takes approximately 5 min, participants are instructed to drive safely in the middle of their lane. Halfway through the drive, participants are instructed to drive at a speed as if they were in a hurry, but that their driving should still remain safely. Hence, there was no specific speed limit in this second half of the drive. The outcome measure of this ride is the Standard Deviation of the Lateral Position, termed Swing-SDLP, that is, the lateral swaying of the car, in the second half of this drive. The Swing-SDLP generally increases as driving speed increases (Dijksterhuis, de Waard, Brookhuis, Mulder, & de Jong, Reference Dijksterhuis, de Waard, Brookhuis, Mulder and de Jong2013). Therefore, in the Swing drive, the Swing-SDLP is related to the amount of risk participants take, because participants are instructed to drive at a speed that is still safe; they need to adjust their driving behavior in order to avoid a risky situation. For instance, they can reduce their speed if they feel that they are losing control. When vehicle control decreases, SDLP increases (Verster & Roth, Reference Verster and Roth2011). The speed at which participants feel comfortable and feel having control, differs from one to another (Dijksterhuis et al., Reference Dijksterhuis, de Waard, Brookhuis, Mulder and de Jong2013). Therefore, driving speed alone would probably be a less accurate measure. A higher Swing-SDLP score indicates increased risk-taking behavior. The Swing-SDLP measures risk-taking behavior without the presence of a direct dangerous situation; a risky situation could be caused by the patients’ behavior.
In the intersections drive, which takes approximately 10 min, participants have to negotiate intersections with varying priority regulations. Participants have to obey the rules, including the speed limits and priority rules. Apart from oncoming traffic on the opposing direction, there is also traffic coming from the right and left at the intersections. At three intersections, participants have to give way. These intersections involve an uncontrolled intersection, a yield-signed intersection, and an intersection with traffic lights. There are two blocks with a maximum speed of 60 km/h and two blocks with a maximum speed of 80 km/h. Participants got one point for each time they did not stop at an intersection and for each speeding violation, based on their average speed per block. Lastly, patients got one point in case of a collision during this ride. The outcome measure, termed Inters-Viol, was the sum score of violations where a higher score indicates more risk-taking behavior. The Inters-Viol measures risk-taking behavior in situations involving anticipating a direct danger, such as a car crossing an intersection.
Statistical Analyses
To test for differences between patients and HCs, t-tests for independent samples for parametric data and Mann–Whitney U (MWU)-tests and Kruskal–Wallis H-tests for non-parametric data were performed. Effect sizes were calculated for the comparisons between patients and HCs (Hedge’s g). Pearson’s correlations were performed to examine the relationship between age and all test variables. To examine the relationship between emotion recognition and risk-taking behavior, partial correlation analyses with age as covariate, between all FEEST scores on the one hand and the AST, Swing-SDLP, and Inters-Viol on the other hand were performed. Missing values in the AST were replaced by the series mean of the corresponding test item. The overall alpha level was set at .05, two-sided. All statistical analyses were performed with IBM SPSS statistics 23 (Armonk, NY, USA).
RESULTS
Missing Values
Regarding the AST, for one patient, three out of 25 items were missing. For one HC, one of the 25 items was missing. Since this concerned < 15% missing, these missing values were replaced by the series mean of the corresponding test item within the respective group (patients or HCs).
Furthermore, eight patients were not able to complete the rides in the driving simulator, due to symptoms of simulator sickness (n = 5, of which three AD, one FTD, and one HD), excessive fatigue (n = 2, of which one FTD and one AD), or inability to comprehend the device instructions (n = 1, DLB). Therefore, Swing-SDLP is based on data from the remaining 43 patients. For Inters-Viol, data are missing for an additional three patients because of simulator sickness (n = 2, both AD) and excessive fatigue (n = 1, DLB), which was not yet present in the Swing drive. Therefore, Inters-Viol is based on the remaining 40 participants. An MWU-analysis showed that patients who were not able to complete the driving simulator drives had significantly lower FEEST-Tot scores in comparison to patients who were able to complete the simulator drives (MWU Z = −2.55, p < .05).
Participants
Descriptive statistics of the three HC groups (HCs who performed the FEEST, HCs who performed the AST, and HCs who performed the driving simulator rides) and the patient group and the patient subgroups are presented in Table 1. There were no significant differences between patients and HCs who performed the FEEST with respect to age (t = −0.15, p =.884), educational level (t = 1.10, p = .271), or sex (χ2 = 2.77, p = .096). Furthermore, there were no significant differences between patients and HCs who performed the driving simulator rides with respect to age (MWU Z = −1.75, p = .080), educational level (MWU Z = −1.23, p = .219), and sex (χ² = .38, p = .538). In addition, there were no significant differences between patients and HCs who performed the AST with respect to sex (χ² = 1.40, p = .236). However, HCs who performed the AST were significantly younger (MWU Z = −2.37, p = .018) and had a significantly higher educational level (MWU Z = −3.46, p = .001). Therefore, age and educational level were entered as covariates for the comparisons between patients and HCs for the AST.
Table 1. Descriptive statistics of the three HC groups, the total patient group, and the subgroups of patients
Note. Education = 7-point scale ranging from 1 (low education) to 7 (university education); HC-FEEST = healthy controls who performed the Ekman 60-faces test of the Facial Expressions of Emotions: Stimuli and Test; HC-AST = healthy controls who performed the Action Selection Test; HC – Driv. Simulator = healthy controls who performed the driving simulator rides.
Table 2 shows that there are no significant differences between the patient subgroups on any of the measures. In addition, the Mini-Mental State Examination (MMSE) score was available for all AD, FTD, and DLB patients and for five HD patients (total n of MMSE scores: 47). A Kruskall–Wallis H-test across all patient subgroups showed that the different subgroups did not statistically differ regarding their MMSE score (H = 6.28, p = .099), indicating that they did not differ regarding their cognitive disease severity.
Table 2. Performance on the FEEST and the AST for all subgroups of patients as well as comparisons between all patient subgroups (Kruskall–Wallis H-tests)
*Chi-value of the chi-square test.
Note. Inters-Viol = risk-taking behavior score of the intersections drive.
Within the patient group, a higher age was significantly associated with a worse perception of anger, that is, a lower FEEST-Ang (r = −.30, p < .05), and with increased risk-taking behavior in the Intersections drive, that is, a higher Inters-Viol (r = .39, p < .05). There were no significant correlations between age and FEEST-Disg (r = −.07, p = .633), FEEST-Fear (r = −.04, p = .784), FEEST-Hap (r = −.25, p = .075), FEEST-Sad (r = −.14, p = .333), FEEST-Sur (r = −.10, p = .487), FEEST-Tot (r = −.21, p = .148), AST (r = .12, p = .419), and Swing-SDLP (r = .28, p = .067).
Emotion Recognition and Risk-Taking Behavior
Table 3 shows that all FEEST scores were significantly lower in patients in comparison to HCs, except for FEEST-Hap and FEEST-Sur. AST and Inters-Viol were significantly higher in the patient group when compared to HCs, that is, patients showed significantly more risk-taking behavior than HCs in the AST and the Intersections drive. There were no significant differences between patients and HCs regarding Swing-SDLP in the second half of the swing drive. However, patients drove significantly slower than HCs in the second half of the swing drive (Swing-Speed). The effect sizes (Hedges’ g) of the significant differences between patients and HCs were moderate to large.
Table 3. Performance on the FEEST, the AST, and the driving simulator (means and SDs) and comparisons between HCs and patients (MWU-tests or ANCOVA for the AST, with age and educational level added as covariates)
* F-value of the ANCOVA instead of the MWU-Z values for the other analyses.
Note. Higher FEEST scores indicate better performance, while a higher AST, Swing-SDLP or Inters-Viol indicate worse performance; ANCOVA = analysis of covariance; Swing-Speed = speed (in km/h) in the swing drive; Inters-Viol = risk-taking behavior score of the intersections drive.
Associations Between Emotion Recognition and Risk-Taking Behavior in the Patient Group
Partial correlations, with age as covariate, between the FEEST and the risk-behavior variables were performed. As can be seen in Table 4, significant correlations were only found for either FEEST-total or FEEST-Fear scores: between FEEST-Tot and AST, between FEEST-Fear and AST and between FEEST-Fear and Inters-Viol. These significant negative correlations indicate that worse performance on the respective FEEST scores was related to increased risk-taking behavior in the AST or Inters-Viol. There were no significant relations between any of the FEEST scores and Swing-SDLP. We found no indications for processing speed to have reduced the scores in the driving simulator, as the time to complete part A of the Trail Making Test was not significantly associated with the Swing-SDLP (r = .01, p = .954) and Inters-Viol (r = .24, p = .141).
Table 4. Partial correlations in the patient group between the FEEST scores and the AST, Swing-SDLP and inters-viol, with age as covariate
Note. Inters-Viol = risk-taking behavior score of the intersections drive.
* p < .05.
DISCUSSION
The main aim of the present study was to investigate whether a neuropsychological test of emotion recognition, an important aspect of social cognition, might be useful to indicate unsafe decision-making behavior in traffic-related situations in a heterogeneous group of patients with NDDs, among which AD, FTD, DLB, and HD. In comparison to HCs, patients took significantly more risks in the AST and in a driving simulator drive. They also performed significantly worse with respect to the overall recognition of facial emotional expressions, and in particular of anger, disgust, fear, and sadness. Moreover, we found that impaired overall recognition of emotional expressions was significantly associated with increased risk-taking behavior in situations involving anticipating a direct danger. Amongst the different basic emotional expressions, only the ability to recognize fearful expressions was found as an indicator of risky decision behavior, which was in line with our expectations. This finding supports the hypothesis that deficits in emotion recognition, an important aspect of social cognition, might be markers of unsafe decision-making behavior in hazardous, traffic-related situations in patients with NDD.
The association between the recognition of particularly fearful facial emotions and risk-taking behavior is thought to be mediated by the underlying ability to experience emotions. According to a prominent theory, the bodily feedback hypothesis, the concepts of facial expression of emotions, experience of emotions, and recognition of emotions are closely related (Price & Harmon-Jones, Reference Price and Harmon-Jones2015). It has been found that experimentally manipulating someone’s emotional expression can influence the experience of that particular emotion (Laird, Reference Laird1974). Furthermore, experimentally blocking someone’s imitation of the facial emotional expression of another individual has been found to interfere with the correct identification of that specific emotion (Niedenthal, Brauer, Halberstadt, & Innes-Ker, Reference Niedenthal, Brauer, Halberstadt and Innes-Ker2001; Oberman, Winkielman, & Ramachandran, Reference Oberman, Winkielman and Ramachandran2007). According to a related theory, the somatic marker hypothesis, the experience of emotions plays, in turn, an important role in decision-making (Damasio et al., Reference Damasio, Tranel and Damasio1991). Each emotion has a specific motivational goal and contributes in a distinctive way to behavior (Zeelenberg, Nelissen, Breugelmans, & Pieters, Reference Zeelenberg, Nelissen, Breugelmans and Pieters2008). In risky situations, the experience of fear is most likely to guide decisions that might have dangerous consequences (Steigenberger, Reference Steigenberger2015). Fearful emotions can be processed quickly, enabling instantaneous responses in order to avoid a risky outcome (Hung et al., Reference Hung, Smith, Bayle, Mills, Cheyne and Taylor2010). Consequently, it seems likely that in the present study patients with a decreased ability to recognize fearful facial expressions, as a probable marker of impaired experience of fear, could not be guided by a fearful experience to prevent them from making risky decisions in the AST and the intersections drive. The finding that only the recognition of fearful expressions, and none of the other basic emotional expressions, was related to risk-taking behavior, suggests that this relationship could not be attributed to an overall decrease in cognition. This idea that emotion recognition is a separate construct is furthermore supported by findings of Spikman et al. (Reference Spikman, Timmerman, Milders, Veenstra and van der Naalt2012), who found that deficits in emotion recognition in patients with TBI were not attributable to deficits in other cognitive functions, such as attention, mental speed, memory, and executive functioning. In addition, our secondary analyses showed that performance on the driving simulator was not confounded by deficits in psychomotor speed. Lastly, the relationship between emotion recognition and risk-taking behavior was controlled for the influence of age.
The significant association between fear recognition and risk-taking behavior was only found in tasks with situations involving direct danger, which was the case in the AST and the intersections drive. In these tasks, various contextual aspects, such as cars or pedestrians suddenly crossing the street, had to be evaluated and anticipated. During this evaluation, patients had to decide what was most appropriate to do in order to avoid a potential dreadful outcome, such as a collision. In the swing drive, on the other hand, patients had to adjust their driving speed in order to decrease the amount of swaying of the car. Not compensating for excessive swaying could have led to serious risky and harmful situations. However, in contrast to the AST and the intersections drive, the contextual aspects of the swing drive did not involve immediate danger. Hence, it seems that the ability to recognize and experience fearful emotions is especially important in guiding decision behavior when the contextual aspects contain direct dangers with potential negative outcomes. These direct dangers probably provoke a fearful emotion, which could in turn guide decision behavior (Schmidt-Daffy, Reference Schmidt-Daffy2013).
The present study included patients with progressive neurological disorders of which it is known that brain networks important for social cognition are likely to be affected in the course of the disease (Elamin et al., Reference Elamin, Pender, Hardiman and Abrahams2012). Previous studies have also included patients with different etiologies on the basis of neuropsychological profiles (Chen et al., Reference Chen, Wells, Otero, Lwi, Haase and Levenson2018; Goodkind et al., Reference Goodkind, Sollberger, Gyurak, Rosen, Rankin, Miller and Levenson2012), which can be methodologically justified (Cipolotti et al., Reference Cipolotti, Healy, Chan, Bolsover, Lecce, White and Bozzali2015). Moreover, separating groups based on a diagnosis of an NDD can be arbitrary, because there can be a large phenotypic heterogeneity within a diagnostic group and a large overlap in symptoms between diagnostic groups (Ryan, Fransquet, Wrigglesworth, & Lacaze, Reference Ryan, Fransquet, Wrigglesworth and Lacaze2018). Indeed, we found no significant differences between the patient subgroups regarding their overall cognition, as assessed with the MMSE, a screening instrument for dementia (Lacy, Kaemmerer, & Czipri, Reference Lacy, Kaemmerer and Czipri2015). In addition, we found that the patient subgroups did not differ regarding their performance on the FEEST, AST, or driving simulator drives.
This study is subject to some limitations. First, several patients in the present study were unable to complete the driving simulator rides (this could not be attributed to a specific disease subgroup), because of simulator sickness or decreased mental endurance, which could have skewed these findings. However, the finding that patients who were not able to complete the driving simulator rides were significantly worse in recognizing emotions than patients who completed the rides, indicates that the current relationships are probably underestimates; based on our findings, those patients with worse emotion recognition abilities would be expected to perform worse in the driving simulator. Classen et al. (Reference Classen, Bewernitz and Shechtman2011) described that simulator sickness is a common problem in especially older people. Fortunately, because of the upcoming technical improvements, including higher refresh rates and better simulator configuration and calibration, simulator sickness is becoming less frequent. Second, the number of HCs who performed the AST and the driving simulator rides was relatively small. However, the main aim of this study was to assess the associations between emotion recognition and risk-taking behavior within the patient group. Lastly, the size of the subgroups of patients was too small to analyze them separately. Consequently, we cannot draw conclusions about differences between the individual etiology subgroups. However, this study was not set up to allow us to examine this issue directly. Moreover, as described earlier, we found no indications for differences in disease severity or differences in test performance between the patient subgroups.
In conclusion, the results of this study show that, irrespective of exact NDD etiology, impaired emotion recognition could be an indicator of daily life risk-taking behavior. This study extends previous studies, which have shown that an impaired emotion recognition can be an indicator of a reduced quality of life and an indicator of the development of behavioral disturbances or mood disorders (Henry, Von Hippel, Molenberghs, Lee, & Sachdev, Reference Henry, Von Hippel, Molenberghs, Lee and Sachdev2016; Phillips, Scott, Henry, Mowat, & Bell, Reference Phillips, Scott, Henry, Mowat and Bell2010). Risk-taking behavior has been found to be an important aspect of the fitness to drive in patients with an NDD (Piersma et al., Reference Piersma, Fuermaier, de Waard, Davidse, de Groot, Doumen and Tucha2016). Therefore, in order to determine whether patients with (early stages of) an NDD are still fit to drive, it is imperative to assess their ability to make safe decisions. However, the assessment of risk-taking behavior directly with the AST or driving simulator rides is hampered by the fact that the availability of these tasks is limited; the AST is not publicly available, because it includes pictures of the theory exam for the Dutch driving license and driving simulators can only be found in few diagnostic settings. Moreover, as described earlier, it took patients time to familiarize with the driving simulator setup. The FEEST does not have these restrictions and can be easily assessed in patients with an NDD. Hence, we suggest that assessing emotion recognition may be a valuable supplement in the assessment of fitness to drive in patients with an NDD. This can be helpful, because finding contributing factors to the fitness to drive in these patients is still a topic of ongoing debate (Jacobs, Hart, & Roos, Reference Jacobs, Hart and Roos2017). Future studies are needed to extend these results by looking at the role of facial emotion recognition in risk-taking behavior in on-road driving behavior and to investigate this relationship in the separate NDD patient groups.
ACKNOWLEDGMENTS
This study was supported by a European Research Council Advanced Grant (#339374) to E. H. F. de Haan.
Conflict of Interest
The authors have nothing to disclose.
Open access
