Cognitive bias modification

A primary focus of our recent work has been to examine the causal link between change in cognitive biases and symptom (e.g., anxiety) reduction using cognitive bias modification paradigms. These training programs are designed to directly alter biased ways of thinking, such as the tendency to make threat interpretations.

For instance, we have trained interpretations to be more benign to decrease anxious responding among obsessional, contamination fearful, socially anxious, trait anxious, spider fearful, and anxiety sensitive samples, and even found that symptom changes following interpretation training for acrophobia (height fear) were as large as those achieved by a group receiving the gold-standard exposure therapy. These demonstrations are significant because they permit evaluation of the causal, rather than simply correlational, claims that underlie cognitive models, and because they offer promise for new interventions that are easy to disseminate given they are technology-based and do not require a therapist.

We launched MindTrails (, a public web site for people to try different online interpretation training programs. Since launching MindTrails, it has been used in more than 85 countries, and various iterations of the intervention have been created, including programs such as MindTrails Spanish (a version of MindTrails adapted for the Latinx community), Hoos Think Calmly (a mobile application version of MindTrails adapted specifically for the UVA community), and MindTrails Teen (a mobile application version of MindTrails adapted for anxious youth). Our goal in developing various web- and mobile-based cognitive bias modification programs is to increase access to evidence-based, mental health interventions in a variety of populations to help overcome common barriers to accessing treatment, such as cost, transportation, and stigma involved in seeking mental health treatment.

Dynamic active and passive sensing of anxiety and emotion regulation

Rather than rely on a static snapshot of biased processing, we seek new ways to more dynamically track anxious and other disorder-related thoughts, feelings, and behaviors. For instance, in collaboration with colleagues in engineering, we use active (e.g., ecological momentary assessment) and passive (e.g., GPS, accelerometer, psychophysiology) mobile sensing via smartphones to learn how social anxiety and depressive symptoms tie to communication patterns and emotion regulation efforts in natural environments. Additionally, these dynamic monitoring approaches can help us identify moments of anxiety in daily life and in turn develop interventions that can be administered right in the moment users need them most.

Difficulty managing emotions puts people at risk for many clinical problems, including anxiety, depression, substance use, and borderline personality disorders. Effective emotion regulation involves a complex sequence of choices in how one responds to emotional challenges in daily life: do you start an emotion regulation (ER) strategy, maintain a strategy, stop a strategy, or switch to a different strategy? This becomes an exponentially complex task when considering that a person has access to many ER strategies and multiple strategies can be used at once. To complicate matters further, fluctuating mood, emotional intensity and demands of the situation all inform strategy choice. Moreover, a person does not make this choice only once in their lifetime; rather, they have to decide how to regulate their emotions over and over again as new stressors unfold. We try to understand emotion regulation decision making, and help anxious people make more adaptive choices.

Implicit associations tied to psychopathology

Implicit associations reflect relatively uncontrollable automatic associations between concepts in memory. Our lab has demonstrated that anxiety and many other forms of psychopathology are characterized by biases in these associations. For instance, we have shown that phobic individuals are more likely than non-phobic individuals to associate their feared object with danger. Moreover, persons with panic disorder associate themselves with panic (vs. calm) relatively more than do nonanxious individuals, and importantly, these associations change following successful treatment and even predict the extent someone will experience a reduction in symptoms.

To allow the public to learn about implicit mental health associations, we direct a public web site called Project Implicit Health ( that allows visitors to try many different Implicit Association Tests tied to their mental and physical health and receive feedback on their score. More than one million tests have been completed!

Attitudes about mental illness and its treatment

Building from the social cognition literature and our understanding of how to modify negative beliefs and attitudes about stigmatized groups, we are examining how automatic biases affect clinical populations. We seek to better understand stigma toward persons with mental illness and treatment seeking, as well as what factors motivate people to seek care (e.g., beliefs about the value of science in determining a treatment plan; role of biases against mentally ill persons at the state level).