Early Emotional Attention is Impacted in Alzheimer’s Disease: An Eye-Tracking Study

Abstract

Emotional deficits have been repetitively reported in Alzheimer’s disease (AD) without clearly identifying how emotional processing is impaired in this pathology. This paper describes an investigation of early emotional processing, as measured by the effects of emotional visual stimuli on a saccadic task involving both pro (PS) and anti (AS) saccades. Sixteen patients with AD and 25 age-matched healthy controls were eye-tracked while they had to quickly move their gaze toward a positive, negative, or neutral image presented on a computer screen (in the PS condition) or away from the image (in the AS condition). The age-matched controls made more AS mistakes for negative stimuli than for other stimuli, and triggered PSs toward negative stimuli more quickly than toward other stimuli. In contrast, patients with AD showed no difference with regard to the emotional category in any of the tasks. The present study is the first to highlight a lack of early emotional attention in patients with AD. These results should be taken into account in the care provided to patients with AD, since this early impairment might seriously degrade their overall emotional functioning.

Keywords

Alzheimer’s disease attention emotion eye movements inhibition

INTRODUCTION

Alzheimer’s disease (AD) is a neurodegenerative pathology involving lesions throughout the brain. The first areas to be affected are the temporal lobes, including the hippocampus [1, 2] and the amygdala [3, 4]. Considering the amygdala’s role in devoting attention to emotional visual stimuli [5 –7] and in extracting emotionally salient features from these stimuli [8 –11], the hypothesis that patients with AD may display an impairment in emotional attention has recently received mounting interest. Yet, to date, the investigations of this particular topic produced contradictory findings. Several studies found that the processing of emotional information is impaired in patients with AD [12 –14], whereas studies of how irrelevant emotional stimuli disturb the processing of non-emotional targets reported that the distractive effect of emotion was unaffected or even enhanced in patients with AD, relative to healthy older adults (HOAs) [15 –17]. The latter findings indirectly suggest that the processing of emotion is unaffected in early-stage AD.

However, several methodological points could explain these discrepancies. Firstly, most of the reported studies used indirect methods that did not feature purely attentional protocols (e.g., working memory tasks involving emotional distractors). To our knowledge, only one study has incorporated a protocol that directly measured selective attention in AD; the results showed that emotional attention was unaffected [18]. Secondly, early versus delayed attentional processes were not solicited to the same extent in the reported studies. We considered that this difference may be crucial for explaining discrepancies between studies, since the amygdala’s activity may primarily sustain early, bottom-up emotional processing [10, 19] based on the modulation of perceptual networks [5, 20] and on automatic orientations toward emotional features [9 , 21–23]. Thus, considering literature reports of early amygdala atrophy in AD and the amygdala’s role in fast, automatic allocation of attentional resources towards emotional information [8 , 20], we hypothesized that patients with AD may display an impairment in early emotional attention. Indeed, the results of studies of attention paid to emotional targets suggest that patients with AD have difficulty in quickly and spontaneously orienting their attention to salient emotional features; in turn, this might produce an impairment in decoding abilities [13] - especially in divided attention paradigms involving visuospatial processing [12]. Decoding abilities can be improved by facilitating orientation toward emotional information [13]; this finding suggests that patients with AD are still able to process emotional information when the latter is presented in low-load attentional conditions (e.g., with fixation or with long presentation times). However, most studies of emotional attention in AD involved complex cognitive processes and thus prevented the assessment of early attentional mechanisms. Hence, in the present study, we used an eye-tracking technique during a simple but time-constrained cognitive task— thereby encouraging the involvement of automatic attentional processes and allowing the accurate measurement of early attentional processes. Eye-tracking paradigms allow access to overt attentional processes [24]. It is generally assumed that visual attention precedes a saccade (eye movement) toward a specific area of the visual field [25]. Moreover, since the fastest saccades occur after around 130 ms, eye movements might be a valuable tool for assessing the time course of attentional orientation and thus probing very early attentional capture.

In classical saccadic tasks, participants can be asked to perform two types of saccades: pro-saccades (PSs) in which they have to saccade as fast as possible toward a target, and anti-saccades (ASs) in which they have to saccade away from the target. Anti-saccades are not ecological phenomena, and are only used as tools for evaluating executive function. Participants first need to inhibit the reflexive saccade toward the peripheral target, and then program and execute a saccade in the opposite direction.

When saccadic tasks use emotional targets, a higher error rate (ER) for emotional stimuli is observed for ASs [26 –29], and faster saccades toward these stimuli as seen in PS-based tasks [26 , 31]. The prioritization of attentional resource allocation toward negative and positive stimuli (relative to neutral information) has been reported in both young adults [32 –34] and HOAs [35 –38]. In early AD, the generation of PSs toward neutral targets is similar to that reported in HOAs [39]. Thus, studying PSs (the primary objective of the present study) enables the investigation of early emotional attention independently of other cognitive processes that might be impaired in AD. As mentioned earlier, ASs are highly demanding in cognitive terms. However, the assessment of AS errors (the secondary objective of the present study) probes early automatic mechanisms, since erroneous saccades (or inhibition errors) correspond to PSs toward the distractor and are driven by exogenous processes [29, 40]. Relative to HOAs, patients with AD make more AS errors toward neutral stimuli [39, 41]; this might reflect impaired inhibition [42]. Hence, the occurrence of error saccades in the AS paradigm enables early emotional capture to be studied. If early emotional attention is unaffected and inhibitory processes are affected in patients with AD, the AS paradigm should enhance the subjects’ sensitivity to emotion - as suggested by previous studies using distracting emotional stimuli in this population [15 –17]. However, if early emotional attention is impaired in AD, emotion should have less impact on AS errors in patients than in HOAs.

Since inhibition is impaired in AD, we expected patients with AD to make more AS errors than HOAs. Given that emotion is known to capture attentional resources, we further expected HOAs to generate more AS errors and show a faster PS for emotional stimuli than for neutral stimuli. In contrast, if early emotional attention is impaired in patients with AD, emotion should have less impact on their saccadic performances (i.e., the PS latency and the AS ER).

MATERIALS AND METHODS

Participants

Fifty-one participants were recruited (18 patients with AD and 33 HOAs). The HOAs were recruited through advertisement on websites and in newspapers. The patients were recruited by the Neurology Unit at Grenoble University Medical Center after a neurological examination, structural magnetic resonance imaging, and a neuropsychological assessment according to the National Institute on Aging and Alzheimer’s Association criteria for probable AD [43]. The mean±standard deviation (SD) score in the Mini-Mental State Examination (MMSE; [44]; French version: [45]) recorded for patients with AD was 24.57±3.41 (range: 18–28). Only HOAs with an MMSE score above 24 (indicating normal cognition [46]) were included (mean±SD MMSE score for the group: 29.28±0.98 (range: 26–30)). The main exclusion criteria were the use of antipsychotic medication, the existence of psychiatric, depressive pathologies, neurological diseases other than AD, a history of brain damage, impaired vision or impaired image processing, inability to understand verbal instructions, and inability to concentrate for the duration of the experiment. All participants had a normal or corrected-to-normal visual acuity. The local investigational review board (Grenoble, France) approved the study. All the participants gave their informed consent prior to study entry, and none received any remuneration.

Two individuals in the AD groups and eight HOAs had to be excluded from the analysis. One patient could not follow the task instructions, and the second was diagnosed with non-amnestic mild cognitive impairment; the diagnosis of AD was not confirmed. The eight HOAs were excluded for one of the following reasons: a high blink rate, drowsiness during the task, calibration issues, a high (≥10) score in the Beck Depression Inventory (BDI-II) [47], and a long reaction time (RT) for PSs (>3 SDs of the group’s mean). The analyzed group of patients with AD (n = 16) comprised nine women and seven men, with a mean±SD age of 74±9 years (range: 57–84). The analyzed group of HOAs (n = 25) comprised 18 women and seven men, with a mean±SD age of 71±7 years (range: 56–83). Demographic and neuropsychological data are summarized in Table 1.

Table 1

Demographic and neuropsychological data for the two groups of participants: patients with Alzheimer’s disease (AD) and healthy older adults (HOAs)

Group	Patients with AD	HOAs
Number	16	25
Women	9 (56.25%)	18 (72%)
Age (y)	74.14±8.57	71.24±7.36
Educational level
Level 1	0%	0%
Level 2	25 %	4%
Level 3	6.25%	12%
Level 4	68.75%	84%
MMSE score	24.57±3.41	29.28±0.98
BDI score	5±3.40	4.02±3.51

Note. Age, educational level, and the MMSE and BDI scores are quoted as the mean±SD. Education level 1: no formal education; level 2: primary education; level 3: secondary education; level 4: high school diploma and above.

There were no significant differences between the HOA group and the AD group in terms of age (t(39) = 2.01, p = 0.16), the BDI score (t(37) = 0.31, p = 0.58), the gender ratio (χ² (1, N = 41) = 1.76, p = 0.18), and the educational level (χ² (2, N = 41) = 4.15, p = 0.13). The two groups differed significantly with regard to the MMSE score (t(38) = 6.68, p < 0.001).

Stimuli

Given that the processing of facial features (e.g., the eyes) is impaired in patients with AD [48], we used emotional visual stimuli other than emotional faces to test the existence of an attention deficit. The stimuli were 96 color images of natural or manufactured objects, presented singly against a white background at a visual angle of 9×9°. A third of the images (i.e., 32) were emotionally positive, a third were emotionally negative, and the last third were emotionally neutral. Performances in trials with neutral stimuli were used as baseline for comparison with performances in trials with negative and positive stimuli. These images were selected from a database of 300 pictures having been rated for valence and arousal by a group of 42 individuals (aged from 24 to 80). This database has already been used in several studies of emotional memory in patients with AD [49 –51]. Each image was independently rated for valence and for arousal on a 1-to-7 scale. Pictures were selected according to the following criteria for valence and arousal: mean ratings for negative valence had to be less than 3, mean ratings for positive valence had to be larger than 5, and mean ratings for neutral pictures had to be between 3.5 and 4.5. The mean ratings for arousal for the three kinds of emotional stimuli had to be between 2 and 6. The same numbers of manufactured and natural images appeared in all groups of images in order to control the degree of semantic similarity between emotional categories.

To control for low-level image features, each picture was characterized by several values computed only for the central object (Fig. 1) using the Python Imaging Library (PIL) module 1 . Brightness corresponded to the image’s mean pixel value when converted into a greyscale. The root mean square contrast was obtained by dividing the standard deviation of the brightness values by the mean brightness of one image (for further details, see [52]). Red, green and blue channel 2 saturation corresponded to the mean pixel color in one image. Pixel complexity was obtained by dividing the number of pixels that were not part of the white background by the total number of pixels. The number of edges gave the number of edge pixels (i.e., points at which the brightness changed sharply or had discontinuities) in one image. The number of lines gave the number of line segments (based on a Hough transform). The images’ characteristics are summarized in Table 2.

Fig.1

Examples of images used as targets to trigger either PS or AS. Top: natural objects Bottom: manufactured objects. From left to right: negative, neutral, and positive stimuli.

Table 2

Characteristics of the three groups of stimuli (quoted as the mean±SD)

Group	Negative	Neutral	Positive
Valence	2.08±1.00	4.13±1.16	5.64±1.10
Arousal	4.27±1.74	2.60±1.59	3.76±1.90
Brightness	124.10±25.44	122.32±23.11	124.76±31.19
RMS contrast	2.08±0.53	2.00±0.37	2.06±0.57
RGB channel saturation
Red	143±31	139±40	160±41
Green	116±25	117±29	108±35
Blue	102±23	97±38	82±33
% pixel complexity	37.16±17.90	33.91±14.58	36.40±11.94
Number of edges	2127±941	1754±810	2003±955
Number of lines	5.22±2.12	4.91±1.30	4.53±1.22
Natural/manufactured ratio	16/16	16/16	16/16

Note. All tests yielded a p-value above 0.05, except for valence (F(2,93) = 1305.55, p < 0.001) and arousal (F(2,93) = 80.27, p < 0.001) tests that revealed stronger arousal for negative images than for positive images (p < 0.001) and for positive images than for neutral images (p < 0.001), and higher valence for positive images than for neutral images (p < 0.001) and for neutral images than for negative images (p < 0.001). RMS: root mean square; RGB: red, green, and blue.

Depending on the task session, the participants had to perform a PS or an AS in response to the target stimulus. The stimuli used in AS tasks were the same as those used in PS tasks. Examples of stimuli are presented in Fig. 1.

Procedure

The experiment was built using SoftEye software [53]. Participants were tested individually, and performed the task on a 21” computer screen with a resolution of 1024×768 pixels and a refresh rate of 100 Hz. Each participant performed the test in a dark room. The participant’s eyes were 64 cm from the computer screen. To avoid movement during the task, the participant’s head was stabilized by a chin-rest and a horizontal forehead support. The participant’s eye movements were recorded with an Eyelink 1000 eye tracker (SR Research, Kanata, ON, Canada) with a time resolution of 500 Hz and a theoretical spatial precision of 0.5°. We used binocular recording and a pupil-corneal reflection mode.

Before the start of the experiment, a 3×3 point calibration sequence was run. Saccades were automatically detected by the Eyelink software using thresholds for velocity (30°/s), acceleration (8000°/s²), and saccadic motion (0.15°).

The experiment comprised four blocks of 48 trials each, separated by a two-minute break. The PS and AS tasks comprised 96 trials each, corresponding to the 96 stimuli selected (3 valence levels×32 different objects). The block order was the same for all participants: one PS block, two AS blocks, and one PS block (based on Antoniades et al.’s standardized protocol [54]). The experiment started with a (less demanding) block of PS, which enabled the participants to familiarize themselves with the task and prepare for the more demanding AS task. The order of the stimuli was randomized. The target location (left or right) was also randomized, so that participants could not predict on which side the target would appear in the next trial. Each block began with a presentation of the task instructions (asking the participant to perform PS or AS), which was followed by 10 training trials and then the recorded session of 48 trials. At the start of each trial, participants had to maintain their gaze on the center of the screen, where a fixation cross appeared for 1600 ms. Next, a target was presented on the left or right side of the screen (at a visual angle of 10° from the center) for 1000 ms. In the PS task, the participants had to direct their gaze toward the target as quickly as possible, whereas in the AS task, they had to direct their gaze away from the target (to the other side of the screen) as quickly as possible. Every 10 trials, the eye-tracking was corrected for drift.

RESULTS

Trials containing blinks were discarded, as were trials with saccades that did not start within 1° of the central fixation cross. The raw data were then filtered to remove anticipation saccades (<80 ms), PSs with a RT above 500 ms, ASs with a RT above 800 ms [55, 56], and trials with saccadic RTs and durations greater than 3 SD for the given subject. This filtering resulted in the exclusion of 19% of the trials. In other words, we assessed a mean±SD of 26±5 useable trials per experimental factor (2 conditions and 3 valence levels). A saccadic response was considered to be correct when the saccade moved toward the target (PS) or away from target (AS) with an amplitude larger than 3°.

Given that emotional stimuli capture attentional resources, we expected HOAs to 1) make more errors for emotional stimuli than for neutral stimuli in the AS task (reflecting early attentional capture) and 2) saccade sooner toward emotional stimuli than toward neutral stimuli in the PS task (reflecting an enhancement of early orientation mechanisms). Lastly, assuming that early emotional attention was impaired in patients with AD, we expected the afore-mentioned effects to be less intense in the AD group. The threshold for statistical significance was set to p < 0.05.

Since the patients with AD and HOAs differed with regard to the gender ratio (which might conceivably have influenced the data), we investigated the influence of gender by including this factor as a covariate in our main analyses. In all analyses, there were no main effects of gender and no interactions between gender and other factors (p > 0.1 for all). Since we had no specific hypotheses regarding gender and given its lack of significant influence, we decided to not to include this variable in our subsequent statistical analyses.

Error rates in the AS task

Error rates in the AS task for HOAs and patients with AD are shown in Fig. 2. We conducted a mixed analysis of variance (ANOVA) on the participants’ ERs, with emotional valence (negative, neutral, positive) as a within-participant factor and group (HOA, AD) as a between-participant factor. Since the variance was not homogeneous, we applied a square root transformation to the raw data [57]. The analysis revealed a significant effect of group (F(1,39) = 24.64, p < 0.001, $η_{p}^{2}$ = 0.39), reflecting a higher ER for patients with AD (62.75±31.69%) than for HOAs (22.65±20.27%). No significant effect of emotional valence (F(2,78) = 1.13, p = 0.33, $η_{p}^{2}$ = 0.03) was observed. There was a trend for the effect of the emotional valence×group interaction (F(2,78) = 2.85, p = 0.06, $η_{p}^{2}$ = 0.07). The planned contrast analysis of differences between HOAs and patients with AD when comparing ERs for emotional and neutral images gave a significant result (b = 0.05, t(78) = 2.04, p < 0.05, 95% confidence interval (CI) [0.001, 0.10], r = 0.23), suggesting that the difference in ER between emotional and neutral images was greater in HOAs (emotional: 23.60±20.51%; neutral: 20.74±20.06%) than in patients with AD (emotional: 62.54±32.54%; neutral: 63.19±30.94%). In view of our starting hypotheses, we performed simple effect analyses for each group in order to further examine the impact of disease on emotional bias. Analysis of ERs for HOAs showed a higher ER for emotional stimuli than for neutral stimuli (b = 0.04, t(76) = 2.61, p < 0.05, 95% CI [0.01, 0.07], r = 0.29). In contrast, analysis of ERs for the patients with AD did not show a significant contrast between emotional conditions (p > 0.1 for all). Bonferroni post hoc tests revealed a higher ER for negative images (24.88±22.18%) than for neutral images (20.74±20.06%, p < 0.05) in HOAs only. Other post hoc contrasts were not significant (p > 0.1 for all).

Fig.2

Mean ER in the AS task as a function of the emotional valence (negative, neutral, and positive) and the study group (HOAs and patients with AD). The error bar represents the corrected standard deviation. * p < 0.05.

Taken together, these data suggest the presence of a higher ER for negative images than for neutral images in HOAs but not in patients with AD.

Saccadic RTs in the AS task

The RT data for HOAs and patients with AD in the AS task are summarized in Table 3. We performed a mixed ANOVA on the mean correct saccadic RTs 3 , with emotional valence (negative, neutral, positive) as a within-participant factor and group (HOA, AD) as a between-participant factor. The main effects of emotional valence (F(2,64) = 0.46, p = 0.64, $η_{p}^{2}$ = 0.01) and of group (F(1,32) = 0.01, p = 0.91, $η_{p}^{2}$ = 0.0004), and the emotional valence×group interaction effect (F(2,64) = 0.42, p = 0.66, $η_{p}^{2}$ = 0.01) were not statistically significant. We also performed a mixed ANOVA on the mean erroneous saccadic RTs 4 , with emotional valence (negative, neutral, positive) as a within-participant factor and group (HOA, AD) as a between-participant factor. The main effects of emotional valence (F(2,54) = 0.10, p = 0.91, $η_{p}^{2}$ = 0.004) and of group (F(1,27) = 0.27, p = 0.61, $η_{p}^{2}$ = 0.010), and the emotional valence×group interaction effect (F(2,54) = 0.10, p = 0.91, $η_{p}^{2}$ = 0.004) were not statistically significant.

Table 3

Mean±SD saccadic RTs in the AS and PS tasks, as a function of the group (patients with AD versus HOAs) and the emotional valence (negative versus neutral versus positive)

Group	Negative	Neutral	Positive
Correct PS RTs
Patients with AD	177±33	182±37	179±29
HOAs	173±24	183±27	178±26
Correct AS RTs
Patients with AD	297±54	303±61	301±51
HOAs	304±67	306±60	299±65
Erroneous AS RTs
Patients with AD	189±37	195±36	192±33
HOAs	187±45	187±34	185±35

Saccadic RTs in the PS task

The RT data for HOAs and patients with AD in the PS task are summarized in Fig. 3 and Table 3. We performed a mixed ANOVA on the mean saccadic RTs5, with emotional valence (negative, neutral, positive) as a within-participant factor and group (HOA, AD) as a between-participant factor. The ANOVA revealed a significant effect of emotional valence (F(2,76) = 6.97, p < 0.05, $η_{p}^{2}$ = 0.16) but not of group (F(1,38) = 0.02, p = 0.89, $η_{p}^{2}$ = 0.0005). The interaction between emotional valence and group was not significant (F(2,76) = 0.81, p = 0.45, $η_{p}^{2}$ = 0.02). The planned contrast analysis of differences between HOAs and patients with AD when comparing ERs for emotional and neutral images gave no significant result (b = 3.80, t(76) = 1.10, p = 0.27, 95% CI [– 5.39, 10.52], r = 0.13). However, given our starting hypotheses, we performed simple effect analyses for each group so as to further examine the impact of disease on emotional bias. Analysis of RTs in HOAs revealed a shorter mean±SD saccadic RT for emotional stimuli (176±25 ms) than for neutral stimuli (183±27 ms, b = 7.21, t(76) = 3.40, p < 0.001, 95% CI [2.99, 11.42], r = 0.36) and a shorter saccadic RT for negative stimuli (173±24 ms) than for positive stimuli (178±26 ms, b = 5.54, t(76) = 2.27, p < 0.05, 95% CI [0.67, 10.41], r = 0.25). In contrast, an analysis of the patients’ RTs did not show any significant contrast between emotional conditions (p > 0.1 for all).

Fig.3

Mean saccadic RT in the PS task as a function of the emotional valence (negative, neutral, and positive) and the study group (HOAs and patients with AD). The error bar represents the corrected standard deviation. * p < 0.05 *** p < 0.001.

Taken as a whole, these results suggest that HOAs have a shorter RT for negative stimuli than for other stimuli. This difference was not observed among patients with AD.

To further explore these data and to estimate the probability of a decline in early orientation toward emotional information in AD, we performed Bayesian analyses on PS RTs [58]. To model the PS RTs as a function of the stimuli’s emotional valence, we used a Bayesian univariate multiple regression model with the PS RT (expressed in ms) as an outcome, valence (negative, neutral, positive) as a within-participant factor, and group (HOA, AD) as a between-participant factor. These analyses were conducted using RStudio [59] and the brms package [60]— an R implementation of Bayesian models that uses the probabilistic programming language Stan6. Four Markov Chain Monte Carlo (MCMC) simulations (referred to as “chains”) were run for each model, including 10,000 iterations and a warmup of 2,000 iterations. Fixed effects were estimated via the posterior mean and 95% highest density intervals (HDIs), where an HDI is the Bayesian analogue of a classical confidence interval. The 95% HDI is an interval that spans 95% of the distribution, such that every point inside the interval has higher believability than any point outside the interval. This strategy allowed us to examine the posterior probability distribution of our parameter of interest (i.e., the difference between HOAs and patients with AD when comparing PS RTs for emotional versus neutral stimuli). The advantage of speed for emotional stimuli versus neutral stimuli was higher for HOAs than for patients with AD (mean [95% HDI] = 3.84 [– 2.89, 10.80]). The corresponding posterior probability distribution is plotted in Fig. 4. The model’s estimation of this contrast is uncertain, as the 95% HDI encompasses zero as a credible value. However, it is noteworthy that 86.5% of the plotted distribution is above zero; in other words, there is a 0.86 probability that the advantage of speed for emotional stimuli versus neutral stimuli is greater in HOAs than in patients with AD.

Fig.4

Histogram of posterior samples of the difference between HOAs and patients with AD when comparing emotional with neutral stimuli, as estimated in a Bayesian univariate multiple regression model.

DISCUSSION

In the present study, we sought to determine the impact of emotional information on early attentional processes in patients with AD. To this end, we used an eye-tracking paradigm and a task that mixed PSs and ASs involving emotional stimuli. On the basis of studies suggesting 1) an impairment in the spontaneous, rapid orientation of attention toward emotional features in AD [13, 62], 2) the existence of early amygdala atrophy in this disease reported in the literature, and 3) the amygdala’s suggested role in fast, automatic allocation of attentional resources towards emotional information [8 , 11], we expected to observe an impairment of early emotional attention in patients with AD (relative to HOAs).

Impairment of early emotional attention in AD

Our experimental data are in accordance with our starting hypotheses, since HOAs showed the early, automatic attentional processing of emotionally negative information (as reflected by the faster generation of PSs and a higher ER in the AS task for this type of stimuli), whereas these results were not observed in a group of age-matched patients with AD. These data suggest that an impairment in early emotional attention can arise in patients with AD under conditions where 1) the emotion is distracting (i.e., the AS task, corresponding to the impairment of attentional capture), and 2) complex cognitive processing is not required, and emotional attention relies on early orientation mechanisms (i.e., the PS task). The saccadic paradigm was particularly appropriate because it examines attentional capture [63, 64] and reduces the involvement of delayed attentional processes in emotional processing [28 , 66]. The limited involvement of delayed attentional processes is a crucial difference with regard to previous studies [15 –17] in which attentional capture by emotional distracting stimuli was preserved when both early and delayed attentional processes could be initiated by patients. In other words, our findings indirectly suggest that preserved performance in emotional-attentional tasks is mainly due to unaffected (and potentially compensating) delayed attentional processes. Our findings agree with reports in which patients had difficulty in quickly and automatically focusing their attention on emotional features of faces [13]. Furthermore, we showed that this impairment can affect emotional items other than human faces.

Given that patients with AD display several cognitive impairments, our results could conceivably be explained by other factors more present in this population than in HOAs (such as inattention, lack of motivation, and forgetfulness). However, the patients were able to execute correct ASs and to correct erroneous ASs - meaning that they had understood and remembered the instructions. Moreover, our patients were recruited in the early stages of AD, as evidenced by the mean±SD MMSE score (24.57±3.41). None of the patients included in the statistical analyses showed any signs of depression, as emphasized by the mean±SD BDI score (5±3.40). Furthermore, the fact that patients with AD and HOAs did not differ in saccadic performance levels (except for AS ERs) suggests that the observed differences were due to selective difficulty in processing emotional information. With regard to the AS ERs, the purpose of the AS task was to determine whether (in a task for which patients with AD are known to be impaired) there was a higher emotional effect on patients than on HOAs, as has been suggested by previous studies [15 –17]. Thus, the fact that the impaired inhibition in patients with AD did not accentuate the emotional effect is in agreement with our hypothesis.

As expected, patients with AD showed a higher overall ER in the AS task than HOAs did, in line with studies showing an impairment of inhibition processes in AD [41, 67]. According to some researchers, this high ER might reflect the impairment of working memory and an instruction omission [68]. However, this theory is controversial [69] and seems unlikely because our patients were able to perform correction saccades. In contrast to previous reports in which decreased inhibition abilities could facilitate the emergence of an emotional effect [15, 17], the patients’ higher ER for ASs did not exacerbate emotional capture. In fact, our data rather suggest the presence of an impairment in early emotional capture in patients with AD, which cannot therefore be enhanced by inhibitory impairments. Conversely, one can assume that this inhibitory impairment would have the opposite effect, i.e., masking the remaining emotional processing in a similar way to a floor effect. Given that we did not observe an effect of emotional valence on RTs in erroneous ASs, this hypothesis nevertheless seems unlikely. Indeed, erroneous ASs are reflexive PSs that occur before they can be inhibited [63]. Even if more stimuli reach the capture threshold in AD [70], emotional stimuli should therefore reach this threshold earlier than neutral stimuli. The absence of this effect suggests a loss of selectivity in AD and a general over-processing of stimuli, regardless of their emotional valence. The observation that patients with AD did not display a difference in RTs for PSs in emotional versus neutral conditions, despite an absence of an impairment in PS generation, also supports this hypothesis.

An alternative explanation for our results relates to the possible alteration of visual perception in AD. On one hand, several studies have suggested that the magnocellular pathway is impaired in AD [71, 72]. This pathway handles the low spatial frequencies (LSFs) [73, 74], which are highly salient in early emotional information processing of images [75, 76]. On the other hand, some studies have suggested that AD is accompanied by a reduction in the focal attention zone [77] and thus in the impaired detection of peripheral elements. Consequently, the influences of LSFs and impaired peripheral detection on emotional attention assessment must be investigated if we are to disentangle these factors from a pure decline in early emotional attention.

Emotional aging: Contribution to socioemotional selectivity theory

Healthy older adults prioritized their processing of emotions, as reflected by a higher ER in the AS task and a shorter RT in the PS task for emotional stimuli than for neutral stimuli. This observation suggests that HOAs’ early attentional processes are sensitive to emotional information. However, planned comparisons and post hoc tests showed that this effect was mainly driven by a significant influence of negative stimuli (rather than positive stimuli) on attentional processing. Our results agree with reports in which the positivity bias in aging corresponds to motivational processing, which is less intense when attention is not fully available [35, 37]. Thus, negative stimuli might automatically capture attention, whereas cognitive control would be required to attend to positive stimuli [78]. Studies of event-related potentials suggest that positivity bias emerges early, yet later than negativity bias [79], meaning that automatic positivity bias may indeed exist in HOAs. However, our eye-tracking protocol was not able to detect this bias because of its heavily time-constrained parameters. Lastly, some researchers suggest that when negative and positive information are moderately arousing, as was the case in the picture database that we used, negative information may preferentially capture attention, especially during early stages of processing [28, 80]. Even though both negative and positive stimuli had moderate arousing ratings in the present study, negative stimuli were slightly more arousing than positive ones. Consequently, to study the influence of arousal on emotional attention, we also analyzed the correlation between arousal ratings for each image and the AS ERs and PS RTs for trials in which the image was presented.

The arousal ratings did not significantly predict the AS ERs in HOAs (r(96) = 0.18, n.s.) or patients with AD (r(96) = – 0.09, n.s.). However, the ratings did significantly predict PS RTs in HOAs (r(96) = – 0.28, p < 0.05; higher image arousal ratings were associated with shorter PS RTs) but not in patients with AD (r(96) = – 0.13, n.s.). The negative correlation between PS RTs and arousal ratings in HOAs was still significant when the variability due to valence ratings was partialed out (semipartial r = – 0.23, p < 0.05). These data are in accordance with research suggesting that early endogenous attention mainly relies on arousal. In other words, PSs (which depend on endogenous saccade generation) could be facilitated by the arousal target’s content independently of its valence [26, 27]. Our data also agree with studies showing that distracting, unpleasant stimuli capture early oculomotor processes more effectively than pleasant stimuli do, especially at moderate levels of arousal [28]. In future studies of the impact of arousal and valence characteristics on attentional mechanisms in normal or pathological aging, it will be important to clearly dissociate between these parameters. However, a major finding of our study is that neither arousal nor valence could explain AS ERs and PS RTs in patients with AD. Our results suggest that early emotional attention is indeed impaired in this population. The mechanisms underlying this impairment remain to be investigated further.

Limitations and perspectives

As shown by our statistical analyses of PS RTs, we were not able to draw firm conclusions about the difference between patients with AD and HOAs when comparing PS RTs for emotional stimuli versus neutral stimuli. Indeed, our frequentist analyses did not lead to a rejection of the null hypothesis, and our Bayesian model estimated that the difference was 3.84 with a broad HDI that included zero as a credible value. Given this result, it would be difficult to adopt a dichotomist decision approach and conclude that the influence of emotion on early processes (measured by the PS RTs) is less pronounced in patients with AD than in HOAs. However, the output of our Bayesian analysis is a posterior probability distribution that can be summarized in many ways. This distribution is plotted on Fig. 4, which also shows the mean [95% HDI] and the proportion of the distribution above and below a particular value (here, we compared the distribution with 0). We found that 86.5% of the distribution is above zero; in other words, there is a 0.86 probability that the advantage of speed for emotional versus neutral stimuli is greater in HOAs than in patients with AD. Even though this estimate suggests that our hypothesis is probable, it needs to be confirmed in further experiments.

The low level of arousal provoked by our data set might explain (at least in part) the absence of an emotional effect in patients with AD. Indeed, poorly arousing stimuli may promote vigilance toward the environment rather than to toward the stimuli themselves. In contrast, strongly arousing stimuli might capture the organism’s attentional resources. Moreover, a number of studies suggest that strong arousal facilitates the preservation of emotional memory [81, 82] and perception abilities in AD [83]. Our results could also be explained by the fact that dimensional models of emotion (which are nevertheless widely applied to normal and pathological populations) are not fully appropriate for studying emotional attention and its impairments. In accordance with appraisal theory, detecting relevance (independently of intrinsic emotion) would enable an organism to establish very quickly whether an event requires the allocation of attentional resources. Indeed, a growing body of literature data shows that personal relevance [32 , 85] and social relevance [86, 87] capture attentional resources. Hence, the assessment of emotional attention with personally relevant, highly arousing stimuli seems necessary for a full investigation of impairments in AD.

The primary objective of the present study was to determine whether early emotional attention mechanisms are affected by AD. Our study’s theoretical framework was based on research suggesting that 1) patients with AD have difficulty in quickly and spontaneously orienting their attention toward emotional features [12, 13], and 2) the amygdala has a role in the fast, automatic allocation of attentional resources toward emotional information [8 , 11] (since atrophy of the amygdala is reportedly a feature of AD). However, we did not directly assess the correlation between emotional attention impairment and atrophy of the amygdala. This step is now essential for characterizing the anatomic sources of the impairment in AD. Recent research suggests that the amygdala is not a single, uniform structure [88]; in fact, it may be an aggregate of sub-nuclei with distinct roles. Notably, atrophy of the centromedial nucleus might impair automatic orientation toward relevant signals in the environment and alter reward conditioning [89], whereas atrophy of the basolateral nucleus might lead to disinhibition of the centromedial region and thus hypervigilance to threat [89 –92] and impaired eye processing when viewing facial expressions [23]. In AD, the initial studies in this field suggested the presence of broad atrophy at the basolateral level [93 –95], although some researchers have suggested that the centromedial nucleus is also affected [93].

Conclusion

To the best of our knowledge, this is the first study to have demonstrated an impairment in early emotional attention to visual stimuli in AD. While arousal and valence had an impact on early attentional mechanisms in HOAs, neither of these factors had a significant influence on saccadic performance in patients with AD. Given that emotional attention influences perception and memory, an impairment of this mechanism might have several consequences on cognitive functioning.

Footnotes

https://www.pythonware.com/products/pil/

Color digital images are made of pixels, which in turn are combinations of the additive primary colors red, green, and blue (RGB). A channel is the grayscale image of the same size as a color image, composed of just one primary color. An RGB image has three channels.

6 patients with AD and 1 HOA were removed from this analysis because they did not provide enough valid data for inclusion (less than 25% useable trials).

2 patients with AD and 10 HOAs were removed from this analysis because they did not provide enough valid data for inclusion (less than 25% useable trials).

Only 15 of the 16 patients with AD were analyzed in this respect because one individual was an outlier (i.e., with inhomogeneous residuals) for a contrast of interest.

Stan implements gradient-based Markov Chain Monte Carlo algorithms, which yield posterior distributions that are straightforward to use for estimating intervals around all parameters.

ACKNOWLEDGMENTS

Funding for this project was provided by a grant from la Région Auvergne Rhône-Alpes. This study was carried out in the Neuropsychology Unit at Grenoble University Hospital, France. We are grateful to all the study participants for kindly giving us their time. We thank Hanna Chainay for the photographs of the emotional stimuli, and Ladislas Nalborczyk for his help on Bayesian analyses. This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Authors’ disclosures available online ().

References

Chetelat

, Baron

(2003) Early diagnosis of Alzheimer’s disease: Contribution of structural neuroimaging. Neuroimage 18, 525–541.

Pennanen

, Kivipelto

, Tuomainen

, Hartikainen

, Hänninen

, Laakso

, Hallikainen

, Vanhanen

, Nissinen

, Helkala

, Vainio

, Vanninen

, Partanen

, Soininen

(2004) Hippocampus and entorhinal cortex in mild cognitive impairment and early AD. Neurobiol Aging 25, 303–310.

Poulin

, Dautoff

, Morris

, Barrett

, Dickerson BC; Alzheimer’s Disease Neuroimaging Initiative (2011) Amygdala atrophy is prominent in early Alzheimer’s disease and relates to symptom severity. Psychiatry Res 194, 7–13.

Klein-Koerkamp

, Heckemann

, Ramdeen

, Moreaud

, Keignart

, Krainik

, Hammers

, Baciu

, Hot P; Alzheimer’s Disease Neuroimaging Initiative (2014) Amygdalar atrophy in early Alzheimer’s disease. Curr Alzheimer Res 11, 239–252.

Vuilleumier

, Richardson

, Armony

, Driver

, Dolan

(2004) Distant influences of amygdala lesion on visual cortical activation during emotional face processing. Nat Neurosci 7, 1271–1278.

Pessoa

, Adolphs

(2010) Emotion processing and the amygdala: From a ‘low road’ to ‘many roads’ of evaluating biological significance. Nat Rev Neurosci 11, 773–783.

Vuilleumier

, Huang

(2009) Emotional attention: Uncovering the mechanisms of affective biases in perception. Curr Dir Psychol Sci 18, 148–152.

Sander

, Grafman

, Zalla

(2003) The human amygdala: An evolved system for relevance detection. Rev Neurosci 14, 303–316.

Jacobs

, Renken

, Aleman

, Cornelissen

(2012) The amygdala, top-down effects, and selective attention to features. Neurosci Biobehav Rev 36, 2069–2084.

10.

Luo

, Holroyd

, Majestic

, Cheng

, Schechter

, Blair

(2010) Emotional automaticity is a matter of timing. J Neurosci 30, 5825–5829.

11.

Diano

, Celeghin

, Bagnis

, Tamietto

(2017) Amygdala response to emotional stimuli without awareness: Facts and interpretations. Front Psychol 7, 2029.

12.

García-Rodríguez

, Vincent

, Casares-Guillén

, Ellgring

, Frank

(2012) The effects of different attentional demands in the identification of emotional facial expressions in Alzheimer’s disease. Am J Alzheimers Dis Other Demen 27, 530–536.

13.

Hot

, Klein-Koerkamp

, Borg

, Richard-Mornas

, Zsoldos

, Paignon Adeline

, Thomas Antérion

, Baciu

(2013) Fear recognition impairment in early-stage Alzheimer’s disease: When focusing on the eyes region improves performance. Brain Cogn 82, 25–34.

14.

Torres

, Santos

, Sousa

, Simões Neto

, Nogueira

, Belfort

, Dias

, Dourado

(2015) Facial expression recognition in Alzheimer’s disease: A longitudinal study. Arq Neuropsiquiatr 73, 383–389.

15.

Doninger

, Bylsma

(2007) Inhibitory control and affective valence processing in dementia of the Alzheimer type. J Neuropsychol 1, 65–83.

16.

Monti

, Weintraub

, Egner

(2010) Differential age-related decline in conflict-driven task-set shielding from emotional versus non-emotional distracters. Neuropsychologia 48, 1697–1706.

17.

Berger

, Erbe

, Ehlers

, Marx

, Hauenstein

, Teipel

(2015) Effects of task-irrelevant emotional stimuli on working memory processes in mild cognitive impairment. J Alzheimers Dis 44, 439–453.

18.

LaBar

, Mesulam

, Gitelman

, Weintraub

(2000) Emotional curiosity: Modulation of visuospatial attention by arousal is preserved in aging and early-stage Alzheimer’s disease. Neuropsychologia 38, 1734–1740.

19.

Méndez-Bértolo

, Moratti

, Toledano

, Lopez-Sosa

, Martínez-Alvarez

, Mah

, Vuilleumier

, Gil-Nagel

, Strange

(2016) A fast pathway for fear in human amygdala. Nat Neurosci 19, 1041–1049.

20.

Rotshtein

, Richardson

, Winston

, Kiebel

, Vuilleumier

, Eimer

, Driver

, Dolan

(2010) Amygdala damage affects event-related potentials for fearful faces at specific time windows. Hum Brain Mapp 31, 1089–1105.

21.

Anderson

, Phelps

(2001) Lesions of the human amygdala impair enhanced perception of emotionally salient events. Nature 411, 305–309.

22.

Murray

, Brosch

, Sander

(2014) The functional profile of the human amygdala in affective processing: Insights from intracranial recordings. Cortex 60, 10–33.

23.

Adolphs

, Tranel

, Damasio

(1994) Impaired recognition of emotion in facial expressions following bilateral damage to the human amygdala. Nature 372, 669–672.

24.

Isaacowitz

, Wadlinger

, Goren

, Wilson

(2006) Is there an age-related positivity effect in visual attention? A comparison of two methodologies. Emotion 6, 511–516.

25.

Rizzolatti

, Riggio

, Dascola

, Umiltá

(1987) Reorienting attention across the horizontal and vertical meridians: Evidence in favor of a premotor theory of attention. Neuropsychologia 25, 31–40.

26.

Kissler

, Keil

(2008) Look-don’t look! How emotional pictures affect pro- and anti-saccades. Exp Brain Res 188, 215–222.

27.

Nummenmaa

, Hyönä

, Calvo

(2009) Emotional scene content drives the saccade generation system reflexively. J Exp Psychol Hum Percept Perform 35, 305–323.

28.

McSorley

, van Reekum

(2013) The time course of implicit affective picture processing: An eye movement study. Emotion 13, 769–773.

29.

Mulckhuyse

, Dalmaijer

(2016) Distracted by danger: Temporal and spatial dynamics of visual selection in the presence of threat. Cogn Affect Behav Neurosci 16, 315–324.

30.

Bannerman

, Milders

, de Gelder

, Sahraie

(2009) Orienting to threat: Faster localization of fearful facial expressions and body postures revealed by saccadic eye movements. Proc Biol Sci 276, 1635–1641.

31.

Schmidt

, Belopolsky

, Theeuwes

(2015) Potential threat attracts attention and interferes with voluntary saccades. Emotion 15, 329–338.

32.

Brosch

, Sander

, Pourtois

, Scherer

(2008) Beyond fear: Rapid spatial orienting toward positive emotional stimuli. Psychol Sci 19, 362–370.

33.

Carretié

, Hinojosa

, López-Martín

, Albert

, Tapia

, Pozo

(2009) Danger is worse when it moves: Neural and behavioral indices of enhanced attentional capture by dynamic threatening stimuli. Neuropsychologia 47, 364–369.

34.

Pool

, Brosch

, Delplanque

, Sander

(2016) Attentional bias for positive emotional stimuli: A meta-analytic investigation. Psychol Bull 142, 79–106.

35.

Knight

, Seymour

, Gaunt

, Baker

, Nesmith

, Mather

(2007) Aging and goal-directed emotional attention: Distraction reverses emotional biases. Emotion 7, 705–714.

36.

Allard

, Isaacowitz

(2008) Are preferences in emotional processing affected by distraction? Examining the age-related positivity effect in visual fixation within a dual-task paradigm. Neuropsychol Dev Cogn B Aging Neuropsychol Cogn 15, 725–743.

37.

Noh

, Isaacowitz

(2015) The effects of varying contextual demands on age-related positive gaze preferences. Psychol Aging 30, 356–368.

38.

Ziaei

, Fischer

(2016) Emotion and Aging. In Neuroimaging Personality, Social Cognition, and Character, Elsevier, Amsterdam, pp.259–278.

39.

Crawford

, Higham

, Renvoize

, Patel

, Dale

, Suriya

, Tetley

(2005) Inhibitory control of saccadic eye movements and cognitive impairment in Alzheimer’s disease. Biol Psychiatry 57, 1052–1060.

40.

Godijn

, Theeuwes

(2002) Programming of endogenous and exogenous saccades: Evidence for a competitive integration model. J Exp Psychol Hum Percept Perform 28, 1039–1054.

41.

Kaufman

, Pratt

, Levine

, Black

(2012) Executive deficits detected in mild Alzheimer’s disease using the antisaccade task. Brain Behav 2, 15–21.

42.

Amieva

, Phillips

, Della Sala

, Henry

(2004) Inhibitory functioning in Alzheimer’s disease. Brain 127, 949–964.

43.

Croisile

, Auriacombe

, Etcharry-Bouyx

, Vercelletto

(2012) Les nouvelles recommandations 2011 du National Institute on Aging et de l’Alzheimer’s Association sur le diagnostic de la maladie d’Alzheimer: Stades précliniques, mild cognitive impairment et démence. Rev Neurol (Paris) 168, 471–482.

44.

Folstein

, Folstein

, McHugh

(1975) “Mini-mental state”. A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res 12, 189–198.

45.

Hugonot-Diener

, Barbeau

, Boivin

, Thomas-Antérion

, Robert

(2008) GREMOIRE: Tests et échelles de la maladie d’Alzheimer et des symptômes apparentés. Solal, Marseille, France.

46.

Mungas

(1991) In-office mental status testing: A practical guide. Geriatrics 46 54–58, 63,66.

47.

Beck

, Steer

, Ball

, Ranieri

(1996) Comparison of Beck Depression Inventories -IA and -II in psychiatric outpatients. J Pers Assess 67, 588–597.

48.

Ogrocki

, Hills

, Strauss

(2000) Visual exploration of facial emotion by healthy older adults and patients with Alzheimer disease. Neuropsychiatry Neuropsychol Behav Neurol 13, 271–278.

49.

Landré

, Sava

, Krainik

, Lamalle

, Krolak-Salmon

, Chainay

(2013) Effects of emotionally-rated material on visual memory in Alzheimer’s disease in relation to medial temporal atrophy. J Alzheimers Dis 36, 535–544.

50.

Chainay

, Sava

, Michael

, Landré

, Versace

, Krolak-Salmon

(2014) Impaired emotional memory enhancement on recognition of pictorial stimuli in Alzheimer’s disease: No influence of the nature of encoding. Cortex 50, 32–44.

51.

Sava

, Paquet

, Krolak-Salmon

, Dumurgier

, Hugon

, Chainay

(2015) Emotional memory enhancement in respect of positive visual stimuli in Alzheimer’s disease emerges after rich and deep encoding. Cortex 65, 89–101.

52.

Bex

, Makous

(2002) Spatial frequency, phase, and the contrast of natural images. J Opt Soc Am A Opt Image Sci Vis 19, 1096–1106.

53.

Ionescu

, Guyader

, Guérin-Dugué

(2009), SoftEye software.

54.

Antoniades

, Ettinger

, Gaymard

, Gilchrist

, Kristjánsson

, Kennard

, John Leigh

, Noorani

, Pouget

, Smyrnis

, Tarnowski

, Zee

, Carpenter

(2013) An internationally standardised antisaccade protocol. Vision Res 84, 1–5.

55.

Malsert

, Grandjean

(2016) Mixed saccadic paradigm releases top-down emotional interference in antisaccade and prosaccade trials. Exp Brain Res 234, 2915–2922.

56.

Noiret

, Vigneron

, Diogo

, Vandel

, Laurent

(2017) Saccadic eye movements: What do they tell us about aging cognition?. Neuropsychol Dev Cogn B Aging Neuropsychol Cogn 24, 575–599.

57.

Howell

(2008) Méthodes statistiques en sciences humaines. De Boeck, Bruxelles, France.

58.

Kruschke

(2015) Doing Bayesian Data Analysis, 2: A Tutorial with R, JAGS, and Stan, Academic Press /Elsevier, Burlington, MA.

59.

RStudio Team (2015) RStudio: Integrated Development for R, RStudio, Inc. Boston, MA.

60.

Bürkner

(2017) brms: An R Package for Bayesian Multilevel Models Using Stan. J Stat Softw 80, doi:.

61.

Carpenter

, Gelman

, Hoffman

, Lee

, Goodrich

, Betancourt

, Brubaker

, Guo

, Li

, Riddell

(2017) Stan: A probabilistic programming language. J Stat Softw 76doi:.

62.

Asaumi

, Morita

, Nakashima

, Muraoka

, Uchimura

(2014) Evaluation of P300 components for emotion-loaded visual event-related potential in elderly subjects, including those with dementia: Visual P300 and dementia. Psychiatry Clin Neurosci 68, 558–567.

63.

Theeuwes

, Kramer

, Hahn

, Irwin

, Zelinsky

(1999) Influence of attentional capture on oculomotor control. J Exp Psychol Hum Percept Perform 25, 1595–1608.

64.

Irwin

, Colcombe

, Kramer

, Hahn

(2000) Attentional and oculomotor capture by onset, luminance and color singletons. Vision Res 40, 1443–1458.

65.

Nummenmaa

, Hyönä

, Calvo

(2006) Eye movement assessment of selective attentional capture by emotional pictures. Emotion 6, 257–268.

66.

Bannerman

, Milders

, Sahraie

(2010) Attentional bias to brief threat-related faces revealed by saccadic eye movements. Emotion 10, 733–738.

67.

Fournet

, Mosca

, Moreaud

(2007) [Deficits in inhibitory processes in normal aging and patients with Alzheimer’s disease: A review]. Psychol Neuropsychiatr Vieil 5, 281–294.

68.

, Ally

(2011) Visual cognition in Alzheimer’s disease and its functional implications. In The Clinical Spectrum of Alzheimer’s Disease - The Charge Toward Comprehensive Diagnostic and Therapeutic Strategies. IntechOpen, London, pp101–126.

69.

Crawford

, Parker

, Solis-Trapala

, Mayes

(2011) Is the relationship of prosaccade reaction times and antisaccade errors mediated by working memory?. Exp Brain Res 208, 385–397.

70.

Carretié

(2014) Exogenous (automatic) attention to emotional stimuli: A review. Cogn Affect Behav Neurosci 14, 1228–1258.

71.

Boucart

, Bubbico

, Szaffarczyk

, Pasquier

(2014) Animal spotting in Alzheimer’s disease: An eye tracking study of object categorization. J Alzheimers Dis 39, 181–189.

72.

Lenoble

, Bubbico

, Szaffarczyk

, Pasquier

, Boucart

(2015) Scene categorization in Alzheimer’s disease: A saccadic choice task. Dement Geriatr Cogn Disord Extra 5, 1–12.

73.

Bullier

(2001) Integrated model of visual processing. Brain Res Brain Res Rev 36, 96–107.

74.

Bar

, Kassam

, Ghuman

, Boshyan

, Schmid

, Dale

, Hämäläinen

, Marinkovic

, Schacter

, Rosen

, Halgren

(2006) Top-down facilitation of visual recognition. Proc Natl Acad Sci U S A 103, 449–454.

75.

Vuilleumier

, Armony

, Driver

, Dolan

(2003) Distinct spatial frequency sensitivities for processing faces and emotional expressions. Nat Neurosci 6, 624–631.

76.

Carretié

, Ríos

, Periáñez

, Kessel

, Alvarez-Linera

(2012) The role of low and high spatial frequencies in exogenous attention to biologically salient stimuli. PLoS One 7, e37082.

77.

Rösler

, Mapstone

, Hays-Wicklund

, Gitelman

, Weintraub

(2005) The “zoom lens” of focal attention in visual search: Changes in aging and Alzheimer’s disease. Cortex 41, 512–519.

78.

Scheibe

, Carstensen

(2010) Emotional aging: Recent findings and future trends. J Gerontol B Psychol Sci Soc Sci 65B, 135–144.

79.

Carretié

, Hinojosa

, Martín-Loeches

, Mercado

, Tapia

(2004) Automatic attention to emotional stimuli: Neural correlates. Hum Brain Mapp 22, 290–299.

80.

Olofsson

, Nordin

, Sequeira

, Polich

(2008) Affective picture processing: An integrative review of ERP findings. Biol Psychol 77, 247–265.

81.

Ikeda

, Mori

, Hirono

, Imamura

, Shimomura

, Ikejiri

, Yamashita

(1998) Amnestic people with Alzheimer’s disease who remembered the Kobe earthquake. Br J Psychiatry 172, 425–428.

82.

Kazui

, Mori

, Hashimoto

, Hirono

, Imamura

, Tanimukai

, Hanihara

, Cahill

(2000) Impact of emotion on memory. Controlled study of the influence of emotionally charged material on declarative memory in Alzheimer’s disease. Br J Psychiatry 177, 343–347.

83.

Elferink

, van Tilborg

, Kessels

(2015) Perception of emotions in mild cognitive impairment and Alzheimer’s dementia: Does intensity matter?. Transl Neurosci 6, 139–149.

84.

Cunningham

, Brosch

(2012) Motivational salience: Amygdala tuning from traits, needs, values, and goals. Curr Dir Psychol Sci 21, 54–59.

85.

Mazzietti

, Sellem

, Koenig

(2014) From stimulus-driven to appraisal-driven attention: Towards differential effects of goal relevance and goal relatedness on attention?. Cogn Emot 28, 1483–1492.

86.

N’Diaye

, Sander

, Vuilleumier

(2009) Self-relevance processing in the human amygdala: Gaze direction, facial expression, and emotion intensity. Emotion 9, 798–806.

87.

Stussi

, Brosch

, Sander

(2015) Learning to fear depends on emotion and gaze interaction: The role of self-relevance in fear learning. Biol Psychol 109, 232–238.

88.

Hortensius

, Terburg

, Morgan

, Stein

, van Honk

, de Gelder

(2017) The basolateral amygdalae and frontotemporal network functions for threat perception. eNeuro 4, doi:.

89.

Mosher

, Zimmerman

, Gothard

(2010) Response characteristics of basolateral and centromedial neurons in the primate amygdala. J Neurosci 30, 16197–16207.

90.

Hortensius

, Terburg

, Morgan

, Stein

, van Honk

, de Gelder

(2017) The dynamic consequences of amygdala damage on threat processing in UrbacheWiethe disease. A commentary on Pishnamazi et al. (2016). Cortex 88, 192–197.

91.

Terburg

, Morgan

, Montoya

, Hooge

, Thornton

, Hariri

, Panksepp

, Stein

, van Honk

(2012) Hypervigilance for fear after basolateral amygdala damage in humans. Transl Psychiatry 2, e115.

92.

de Gelder

, Terburg

, Morgan

, Hortensius

, Stein

, van Honk

(2014) The role of human basolateral amygdala in ambiguous social threat perception. Cortex 52, 28–34.

93.

Cavedo

, Pievani

, Boccardi

, Galluzzi

, Bocchetta

, Bonetti

, Thompson

, Frisoni

(2014) Medial temporal atrophy in early and late-onset Alzheimer’s disease. Neurobiol Aging 35, 2004–2012.

94.

Tang

, Holland

, Dale

, Younes

, Miller MI; Alzheimer’s Disease Neuroimaging Initiative (2015) The diffeomorphometry of regional shape change rates and its relevance to cognitive deterioration in mild cognitive impairment and Alzheimer’s disease. Hum Brain Mapp 36, 2093–2117.

95.

Ortner

, Pasquini

, Barat

, Alexopoulos

, Grimmer

, Förster

, Diehl-Schmid

, Kurz

, Förstl

, Zimmer

, Wohlschläger

, Sorg

, Peters

(2016) Progressively disrupted intrinsic functional connectivity of basolateral amygdala in very early Alzheimer’s disease. Front Neurol 7, 132.