Individual Differences In Anxiety Research Articles

Computational modeling of social evaluative decision-making elucidates individual differences in adolescent anxiety.

Adolescents experience significant developmental changes during a time of heightened sensitivity to social cues, particularly rejection by peers, which can be especially overwhelming for those with elevated levels of social anxiety. Social evaluative decision-making tasks have been useful in uncovering the neural correlates of information processing biases; however, linking youths' task-based performance to individual differences in psychopathology (e.g., anxiety symptoms) has proven more elusive. Here, we address this weakness with drift diffusion modeling to decompose youths' performance on the social judgment paradigm (SJP) to determine if this approach is useful in discovering individual differences in anxiety symptoms, as well as puberty, age, and sex. A sample of 103 adolescents (55 males, Mage = 14.49, SD = 1.69) completed the SJP and self-report measures of anxiety, as well as self- and parent-reported measures of puberty. The decision threshold parameter, reflecting the amount of evidence needed to make a social evaluative decision, predicted youth self-reported anxiety, above and beyond typical metrics of SJP performance. Our results highlight the potential advantage of parsing task performance according to the underlying cognitive processes. Future research would likely benefit from applying computational modeling approaches to social judgment tasks when attempting to uncover performance-based individual differences in psychopathology.

Smile, You’re on Camera: Investigating the Relationship between Selfie Smiles and Distress

Background: This study examined the relationship between (1) participant smiling in daily “selfie” videos and (2) self-reported distress. Given the extensive use of digital devices for sharing expressions of non-verbal behavior, and some speculation that these expressions may reveal psychological states—including emotional distress—we wanted to understand whether facial expression in these TikTok-like videos were correlated with standardized measures of psychological distress. Based on the work of Paul Ekman and others, which posits that facial expressions are universal reflections of people’s inner states, we predicted that smiling would be inversely related to psychological distress. Method: Twenty-four undergraduate students, aged 18+ years (M = 18.35, SD = 2.75), were prompted to record a two-minute selfie video each evening during two weeks of data collection (i.e., 14 total days). They were instructed to describe various aspects of their day. They also completed self-report questionnaires at the end of each assessment week, including the Depression Anxiety Stress Scale (DASS), Perceived Stress Scale (PSS), and the Pittsburgh Sleep Quality Index (PSQI). Results: A counterintuitive effect was observed whereby smiling intensity during selfie videos was positively correlated with individual differences in anxiety, depression, and stress. Discussion: This study challenges the common view that facial expressions necessarily reflect our inner emotions. It provides preliminary evidence that a mobile sensing app that captures selfies—along with other naturalistic data—may help elucidate the relationship between facial expressions and emotions.

Sex differences in the effects of trait anxiety and age on resting-state functional connectivities of the amygdala

BackgroundNumerous studies characterized how resting-state functional connectivities (rsFCs) of the amygdala were disrupted in emotional disorders and varied with emotional traits, including anxiety. With trait anxiety known to diminish with age, a critical issue concerns disambiguating the effects of age and anxiety on amygdala rsFCs in studying the neural bases of individual differences in anxiety. MethodsTwo-hundred adults (83 women) 19–85 years of age underwent fMRI and assessment for trait anxiety. Amygdala rsFC correlates were identified using multiple regression with age and anxiety in the same model for all and separately in men and women. The rsFC correlates were examined for age-anxiety interaction. ResultsAnxiety was negatively correlated with amygdala-temporooccipital gyri rsFC in all and in men alone. In women, amgydala rsFC with the thalamus/pallidum, angular/supramarginal gyri, inferior temporal gyrus, and posterior insula correlated positively and rsFC with calcarine cortex and caudate correlated negatively with anxiety. We also observed sex differences in age correlation of amgydala-posterior cingulate cortex/precuneus and -insula/temporoparietal rsFCs, with stronger associations in women. In women alone, anxiety and age interacted to determine amygdala rsFC with the thalamus/pallidum, calcarine cortex, and caudate, with older age associated with stronger correlation between anxiety and the rsFCs. LimitationsThe findings need to be validated in an independent sample and further explored using task-based data. ConclusionHighlighting anxiety- and age- specific as well as interacting correlates of amygdala rsFCs and sex differences in the correlates, the findings may shed light on the neural markers of anxiety.

Reported Use of Second-Language Speech Accommodation in Everyday Interactions: The Role of Individual Differences.

Proficient speakers of a language often accommodate less proficient speakers during conversation to facilitate comprehension, but information about factors such as personality and language experience that may shape how speakers perceive accommodation is limited. We developed an online questionnaire to clarify the use of speech accommodation in relation to individual differences in anxiety, personality, and English proficiency. Using Qualtrics Panels for recruitment, we surveyed a representative sample of second-language (L2) English speakers (n = 201) and first-language (L1) English speakers (n = 192) across the United States. We report descriptive results in addition to correlations and a factor analysis to assess the perception of accommodation in L2 and L1 speakers. Only a third of L2 participants reported that L1 speakers change their speech when talking to them, and more than half are frustrated when L1 speakers do not accommodate them. Indeed, a majority of our L1 participants reported that they do not change their speech when talking to L2 speakers. For both groups, measures of anxiety, personality, and L2 proficiency modify results, providing novel evidence on factors that influence L2 accommodation. Results suggest that L1 speakers accommodate L2 speakers less frequently than previously reported. The data are discussed under communication accommodation theory.

Examining a model of anxiety in autistic adults.

Anxiety disorders commonly occur in autism. Existing studies implicate intolerance of uncertainty, alexithymia, sensory processing differences and emotion regulation difficulties as influencing factors of anxiety in autism. To date, a few studies have considered the combination of these factors within the same sample. This study used structural equation modelling to test the prediction that intolerance of uncertainty and emotion regulation constitute more direct causes of anxiety in autism that mediate the influences of sensory processing difference and alexithymia as more sequential contributing factors. Autistic (n = 86) and non-autistic adults (n = 100) completed a battery of self-report questionnaires. Only when applied to each group separately, the broad predictions of the model were confirmed for the autistic group following data-driven additions of paths between sensory processing difference and anxiety and alexithymia implying that sensory processing difference contribute indirectly as well as directly to individual differences in anxiety. For the non-autistic group, model fit could only be achieved after removing autism-related traits and sensory processing differences as predictors of anxiety. These results suggest that aetiology and expression of anxiety in autism partially overlap with what is observed in the general population except that sensory processing differences appear to play a relatively unique role in the context of autism.

Trait Anxiety Influences Negative Affect-modulated Distribution of Visuospatial Attention

Visuospatial attention allows humans to selectively gate and prioritize visual (including salient, emotional) information for efficiently navigating natural visual environments. As emotions have been known to influence attentional performance, we asked if emotions also modulate the spatial distribution of visual attention and whether any such effect was further associated with individual differences in anxiety. Participants (n = 28) discriminated the orientation of target Gabor patches co-presented with distractors, speedily and accurately. The key manipulation was randomly presenting a task-irrelevant face emotion prime briefly (50 ms), conveying Neutral/Disgust/Scrambled (Null) emotion signal 150 ms preceding the target patches. We calculated attention gradient (change in negative inverse attentional efficiency with unit change in distance from the source of emotion signal) as a metric to answer our questions. Specifically, the Disgust signal modulated the direction of attention gradients differentially in individuals with varying degrees of trait - anxiety, such that the gradients correlated negatively with individual trait-anxiety scores. This implies spatial shifts in Disgust-signalled visual attention with varying trait - anxiety levels. Neutral yielded attention gradients comparable to Scrambled, implying no specific effect of this signal and there was no association with anxiety levels in both. No correlation was observed between state - anxiety and the emotion-cued attention gradients. In sum, the results suggest that individual trait - anxiety levels influence the effect of negative and physiologically arousing emotion signals (e.g., Disgust) on the spatial distribution of visual attention. The findings could be of relevance for understanding biases in visual behaviour underlying affective states and disorders.

White-matter correlates of anxiety: The contribution of the corpus-callosum to the study of anxiety and stress-related disorders.

Traumatic stress has been associated with increased risk for brain alterations and development of anxiety disorders. Studies conducted in posttraumatic patients have shown white-mater volume and diffusion alterations in the corpus-callosum. Decreased cognitive performance has been demonstrated in acute stress disorder and posttraumatic patients. However, whether cognitive alterations result from stress related neuropathology or reflect a predisposition is not known. In the current study, we examined in healthy controls, whether individual differences in anxiety are associated with those cognitive and brain alterations reported in stress related pathologies. Twenty healthy volunteers were evaluated for anxiety using the state-trait inventory (STAI), and were tested for memory performance. Brain imaging was employed to extract volumetric and diffusion characteristics of the corpus-callosum. Significant correlations were found between trait anxiety and all three diffusion parameters (fractional-anisotropy, mean and radial-diffusivity). Associative-memory performance and corpus-callosum volume were also significantly correlated. We suggest that cognitive and brain alterations, as tested in the current work and reported in stress related pathologies, are present early and possibly persist throughout life. Our findings support the hypothesis that individual differences in trait anxiety predispose individuals towards negative cognitive outcomes and brain alterations, and potentially to stress related disorders.

Anxiety and Avoidance in Close Relationships Are Associated With Responses to Negative and Positive Stimuli.

Humans tend to assign valence to objects, people, and events in the environment, but there are individual differences in the evaluation of the affective nature of these environmental stimuli. This exploratory study investigated how individual differences in anxiety and avoidance in close relationships are associated with the emotional appraisal of valenced and neutral stimuli. Participants evaluated negative, neutral, and positive stimuli for emotional valence in an image classification task. There was a positivity offset across all participants, in that neutral stimuli were evaluated as more positive than negative. Individuals higher on the Experiences in Close Relationships-Anxiety subscale showed a negativity bias in reaction times and ratings: they had faster response times to negative than to positive stimuli and had a greater tendency to evaluate positive stimuli as "negative." Individuals higher on the Experiences in Close Relationships-Avoidance subscale gave more positive ratings of negative stimuli and more negative ratings of positive stimuli, which may suggest a general blunted response to emotional stimuli. Findings are discussed in the context of the literature on individual differences and emotional appraisal of stimuli.

Fear influences phantom sound percepts in an anechoic room

Aims and hypothesesIn an environment of absolute silence, researchers have found many of their participants to perceive phantom sounds (tinnitus). With this between-subject experiment, we aimed to elaborate on these research findings, and specifically investigated whether–in line with the fear-avoidance model of tinnitus perception and reactivity–fear or level of perceived threat influences the incidence and perceptual qualities of phantom sound percepts in an anechoic room. We investigated the potential role of individual differences in anxiety, negative affect, noise sensitivity and subclinical hearing loss. We hypothesized that participants who experience a higher level of threat would direct their attention more to the auditory system, leading to the perception of tinnitus-like sounds, which would otherwise be subaudible, and that under conditions of increased threat, narrowing of attention would lead to perceptual distortions.MethodsIn total, N = 78 normal-hearing volunteers participated in this study. In general, the study sample consisted of young, mostly female, university students. Their hearing was evaluated using gold-standard pure tone audiometry and a speech-in-noise self-test (Digit Triplet Test), which is a sensitive screening test to identify subclinical hearing loss. Prior to a four-minute stay in an anechoic room, we randomized participants block design-wise in a threat (N = 37) and no-threat condition (N = 41). Participants in the threat condition were deceived about their hearing and were led to believe that staying in the room would potentially harm their hearing temporarily. Participants were asked whether they perceived sounds during their stay in the room and rated the perceptual qualities of sound percepts (loudness and unpleasantness). They were also asked to fill-out standardized questionnaires measuring anxiety (State–Trait Anxiety Inventory), affect (Positive and Negative Affect Schedule) and noise sensitivity (Weinstein Noise Sensitivity Scale). The internal consistency of the questionnaires used was verified in our study sample and ranged between α = 0.61 and α = 0.90.ResultsIn line with incidence rates reported in the literature, 74% of our participants reported having heard tinnitus-like sounds in the anechoic room. Speech-in-noise identification ability was comparable for both groups of participants. The experimental manipulation of threat was proven to be effective, as indicated by significantly higher scores on a Threat Manipulation Checklist among participants in the threat condition as compared to those in the no-threat condition (p < 0.01). Nevertheless, participants in the threat condition were as likely to report tinnitus percepts as participants in the no-threat condition (p = 1), and tinnitus percepts were not rated as being louder (p = 0.76) or more unpleasant (p = 0.64) as a function of level of threat. For participants who did experience tinnitus percepts, a higher level of threat was associated with a higher degree of experienced unpleasantness (p < 0.01). These associations were absent in those who did not experience tinnitus. Higher negative affect was only slightly associated with higher ratings of tinnitus unpleasantness (p < 0.01).ConclusionWhereas our threat manipulation was successful in elevating the level of fear, it did not contribute to a higher percentage of participants perceiving tinnitus-like sounds in the threat condition. However, higher levels of perceived threat were related to a higher degree of perceived tinnitus unpleasantness. The findings of our study are drawn from a rather homogenous participant pool in terms of age, gender, and educational background, challenging conclusions that are applicable for the general population. Participants generally obtained normophoric scores on independent variables of interest: they were low anxious, low noise-sensitive, and there was little evidence for the presence of subclinical hearing loss. Possibly, there was insufficient variation in scores to find effects.

Construction and Validation of a Scale to Measure Loneliness and Isolation During Social Distancing and Its Effect on Mental Health.

A variety of factors contribute to the degree to which a person feels lonely and socially isolated. These factors may be particularly relevant in contexts requiring social distancing, e.g., during the COVID-19 pandemic or in states of immunodeficiency. We present the Loneliness and Isolation during Social Distancing (LISD) Scale. Extending existing measures, the LISD scale measures both state and trait aspects of loneliness and isolation, including indicators of social connectedness and support. In addition, it reliably predicts individual differences in anxiety and depression. Data were collected online from two independent samples in a social distancing context (the COVID-19 pandemic). Factorial validation was based on exploratory factor analysis (EFA; Sample 1, N = 244) and confirmatory factor analysis (CFA; Sample 2, N = 304). Multiple regression analyses were used to assess how the LISD scale predicts state anxiety and depression. The LISD scale showed satisfactory fit in both samples. Its two state factors indicate being lonely and isolated as well as connected and supported, while its three trait factors reflect general loneliness and isolation, sociability and sense of belonging, and social closeness and support. Our results imply strong predictive power of the LISD scale for state anxiety and depression, explaining 33 and 51% of variance, respectively. Anxiety and depression scores were particularly predicted by low dispositional sociability and sense of belonging and by currently being more lonely and isolated. In turn, being lonely and isolated was related to being less connected and supported (state) as well as having lower social closeness and support in general (trait). We provide a novel scale which distinguishes between acute and general dimensions of loneliness and social isolation while also predicting mental health. The LISD scale could be a valuable and economic addition to the assessment of mental health factors impacted by social distancing.

Anxiety and amygdala connectivity during movie-watching

Rodent and human studies have implicated an amygdala-prefrontal circuit during threat processing. One possibility is that while amygdala activity underlies core features of anxiety (e.g. detection of salient information), prefrontal cortices (i.e. dorsomedial prefrontal/anterior cingulate cortex) entrain its responsiveness. To date, this has been established in tightly controlled paradigms (predominantly using static face perception tasks) but has not been extended to more naturalistic settings. Consequently, using ‘movie fMRI’—in which participants watch ecologically-rich movie stimuli rather than constrained cognitive tasks—we sought to test whether individual differences in anxiety correlate with the degree of face-dependent amygdala-prefrontal coupling in two independent samples. Analyses suggested increased face-dependent superior parietal activation and decreased speech-dependent auditory cortex activation as a function of anxiety. However, we failed to find evidence for anxiety-dependent connectivity, neither in our stimulus-dependent or -independent analyses. Our findings suggest that work using experimentally constrained tasks may not replicate in more ecologically valid settings and, moreover, highlight the importance of testing the generalizability of neuroimaging findings outside of the original context.

The Role of Student Debt and Debt Anxiety in College Student Financial Well-Being

This study examined predictors of three measures of financial well-being in 354 college students. Results suggest that perceptions of debt are important in understanding financial well-being, but these perceptions need to be considered alongside of individual differences in anxiety, optimism, and parental teaching. Further, regression analyses highlighted differences between predictors of perceived financial well-being and more and less frequent money saving behaviors, which may indicate different paths for intervention. Specifically, students reporting engaging in less common financial strain behaviors appear to have more objective financial difficulties, whereas students reporting more common financial strain behaviors report more factors that suggest difficulty with money management skills and those with lower levels of perceived financial well-being report higher levels of worry. Mediational analyses further indicated a potential role for parental intervention to improve perceived financial well-being and decrease unhealthy responses to financial strain.

They're watching you: the impact of social evaluation and anxiety on threat-related perceptual decision-making.

In day-to-day social interactions, we frequently use cues and contextual knowledge to make perceptual decisions regarding the presence or absence of threat in facial expressions. Such perceptual decisions are often made in socially evaluative contexts. However, the influence of such contexts on perceptual discrimination of threatening and neutral expressions has not been examined empirically. Furthermore, it is unclear how individual differences in anxiety interact with socially evaluative contexts to influence threat-related perceptual decision-making. In the present study, participants completed a 2-alternative forced choice perceptual decision-making task in which they used threatening and neutral cues to discriminate between threatening and neutral faces while being socially evaluated by purported peers or not. Perceptual sensitivity and reaction time were measured. Individual differences in state anxiety were assessed immediately after the task. In the presence of social evaluation, higher state anxiety was associated with worse perceptual sensitivity, i.e., worse discrimination of threatening and neutral faces. These findings suggest that individual differences in anxiety interact with social evaluation to impair the use of threatening cues to discriminate between threatening and neutral expressions. Such impairment in perceptual decision-making may contribute to maladaptive social behavior that often accompanies evaluative social contexts.

Associations Between Anxiety Symptoms and Health-Related Quality of Life: A Population-Based Twin Study in Sri Lanka

Anxiety not only concerns mental wellbeing but also negatively impacts other areas of health. Yet, there is limited research on (a) the genetic and environmental aetiology of such relationships; (b) sex differences in aetiology and (c) non-European samples. In this study, we investigated the genetic and environmental variation and covariation of anxiety symptoms and eight components of health-related quality of life (QoL), as measured by the short form health survey (SF-36), using genetic twin model fitting analysis. Data was drawn from the Colombo Twin and Singleton Study (COTASS), a population-based sample in Sri Lanka with data on twins (N = 2921) and singletons (N = 1027). Individual differences in anxiety and QoL traits showed more shared environmental (family) effects in women. Men did not show familial effects. Anxiety negatively correlated with all eight components of QoL, mostly driven by overlapping unique (individual-specific) environmental effects in both sexes and overlapping shared environmental effects in women. This is the first study in a South Asian population supporting the association between poor mental health and reduced QoL, highlighting the value of integrated healthcare services. Associations were largely environmental, on both individual and family levels, which could be informative for therapy and intervention.

Low working memory load facilitates attention bias modification training

Implementations of attention bias modification training (ABMT) attempt to retrain attention away from rather than towards threat, thereby disrupting the anxiety-related attentional bias (AB). Yet, results of ABMT studies have been mixed due to limitations in knowledge of mechanisms underlying ABMT efficacy. Dual-process models of anxiety posit that ABMT works primarily through strengthening of the top-down cognitive control of attention to threat. If this is the case, introducing a working memory load (WML) during ABMT should reduce training efficacy. However, prior studies employing this method show mixed results (Booth, Mackintosh, Mobini, Oztop, & Nunn, 2014; Clarke et al., 2017) and fail to directly compare low and high WML with no WML or to account for individual differences in anxiety and working memory capacity (WMC). The present study (N = 306) assessed trait anxiety and WMC in neurotypical adults who were then randomly assigned to ABMT that trained attention toward or away from threat, with either no, low, or high WML, for a total of six training groups. Attentional bias was assessed before and after training. Results showed ABMT successfully trained attention under low WML, but not under high or no WML, suggesting that ABMT is facilitated by engaging but not overtaxing WML.

Expressive flexibility and anxiety

BackgroundThe ability to flexibly enhance and suppress emotional expressions plays an important role in emotion regulation and the fostering of social connections. Deficits in expressive flexibility have been linked with posttraumatic stress disorder and prolonged grief reactions. Previous studies have suggested that the ability to respond flexibly in the context of an immediate threat (a nonconscious prime) may be important. In this study we sought to extend this work by examining the impact of individual differences in anxiety on expressive flexibility. MethodsEighty four participants with high and low trait anxiety viewed blocks of negative images with instructions to enhance or suppress emotional expressions. Across blocks one of three nonconscious primes (threat, safety, neutral) appeared before each image. Observers blind to the study's design rated participants’ emotional expressiveness. ResultsRepeated Measures Anova's showed that high trait anxiety was associated with lower levels in enhancement ability. Further, low trait anxiety was associated with less emotion in response to the threat prime irrespective of expressive instruction. LimitationsThis study was cross-sectional which precludes causal relationships. Participants were drawn from a student population and the generalizability to other populations will need to be established ConclusionsThis study identified trait anxiety as factor that may contribute to expressive flexibility deficits. High trait anxiety is a feature of many clinical conditions. Strategies to target expressive flexibility may relevant to a range of anxiety related clinical conditions.

Anxiety, Depression, and Opioid Misuse Among Adults With Chronic Pain: The Role of Anxiety Sensitivity.

The opioid epidemic is a significant public health problem largely driven by opioid prescriptions for chronic pain. Among those with chronic pain, anxiety and depressive symptoms have been linked to opioid misuse, and individual differences in anxiety and depressive symptoms among adults with chronic pain may be important for better understanding pain. Yet, little work has examined mechanisms that may link anxiety and depressive symptoms to opioid misuse among adults with chronic pain. Anxiety sensitivity (AS), or the fear of anxiety-related physical sensations, may be one candidate construct that has been linked independently to anxiety and depressive symptoms as well as opioid misuse. The current survey-based study examined the indirect association of AS in the relation between anxiety and depressive symptoms and opioid misuse among 429 adults with chronic pain currently using prescription opioid medication (73.9% female, Mage=38.32 y, SD=11.07). Using structural equation modeling, it appears that AS is a potential construct in the interrelation between anxiety and depressive symptoms and opioid misuse. However, an alternative model found equally strong empirical support, as mental health symptoms may serve as an explanatory factor between AS and opioid misuse. Bi-directional effects are apt to be involved among the studied variables. Prospective research is needed to replicate the study results and isolate the temporal patterning between the studied constructs. Nonetheless, the data overall highlight the importance of both AS and anxiety and depressive symptoms in terms of understanding opioid misuse among adults with chronic pain.

Approach and avoidance patterns in reward learning across domains: An initial examination of the Social Iowa Gambling Task

The current study examines learning patterns in response to both monetary and social incentives through both approach and avoidance behaviors using modified versions of the Iowa Gambling Task. Specifically, we investigated learning in response to both positive and negative feedback in a sample of 191 undergraduate students. The social task was a novel paradigm, and social feedback were images of faces displaying positive and negative emotions. We examined internal validity of the tasks through modeling changes in approach and avoidance. We also explored associations between approach and avoidance learning and individual differences in anxiety and social anxiety, depression and well-being, general anhedonia and social closeness, and fun-seeking, using multilevel models (MLMs). Results showed that both the monetary and social tasks demonstrated learning as shown by decreases in plays on disadvantageous decks across the task. Additionally, we found that overall task performance on the monetary task was associated with fun-seeking and overall task performance on the social task was associated with fun-seeking and depressive symptoms. Initial findings suggest promise for the novel task in the examination of social avoidance learning.

Individual differences in anxiety and automatic amygdala response to fearful faces: A replication and extension of Etkin et al. (2004)

Trait anxiety refers to the stable tendency to attend to threats and experience fears and worries across many situations. According to the widely noticed, pioneering investigation by Etkin et al. (2004) trait anxiety is strongly associated with reactivity in the right basolateral amygdala to non-conscious threat. Although this observation was based on a sample of only 17 individuals, no replication effort has been reported yet. We reexamined automatic amygdala responsiveness as a function of anxiety in a large sample of 107 participants. Besides self-report instruments, we administered an indirect test to assess implicit anxiety. To assess early, automatic stages of emotion processing, we used a color-decision paradigm presenting brief (33 ms) and backward-masked fearful facial expressions. N = 56 participants were unaware of the presence of masked faces. In this subset of unaware participants, the relationship between trait anxiety and basolateral amygdala activation by fearful faces was successfully replicated in region of interest analyses. Additionally, a relation of implicit anxiety with masked fear processing in the amygdala and temporal gyrus was observed. We provide evidence that implicit measures of affect can be valuable predictors of automatic brain responsiveness and may represent useful additions to explicit measures. Our findings support a central role of amygdala reactivity to non-consciously perceived threat in understanding and predicting dispositional anxiety, i.e. the frequency of spontaneously occurring anxiety in everyday life.

Individual differences in anxiety are related to differences in learning performance and cognitive style

Individuals of the same population differ in their cognitive abilities, that is, in their capacity to acquire, process, store and act on information from the environment. The question of whether individuals with different personality types differ in their learning performances has been commonly addressed but has yielded mixed results. We propose that methodological differences among studies might have contributed to these contrasting results. Notably, the strength and direction of the association between personality and learning might be affected by which personality traits are considered and how they are measured. To test this hypothesis, we conducted an experiment with sailfin mollies, Poecilia latipinna. We assessed three personality traits that are likely to affect individuals’ performance by influencing either their willingness to encounter the new situation (i.e. exploration and neophobia) or their susceptibility or emotional reactivity to stressful situations (i.e. anxiety). Then, we scored their performance in two learning tasks (i.e. discrimination and spatial reversal learning) after they had been extensively familiarized with the learning device and hence were all willing to encounter the task. We found that only anxiety had a significant effect on learning performance, with less anxious fish performing better in the discrimination learning task but worse in the spatial reversal task than highly anxious ones. Thus, our findings confirm a link between personality and cognition and are consistent with the idea that learning requires different steps that are each associated with different personality traits. Furthermore, highly anxious fish, on average, needed more trials to learn an association but some of them (with presumably high energetic requirements) consistently took more time before making a choice and had higher performance than their counterparts that made faster decisions.