Christopher D. Hornig’s research while affiliated with Harvard University and other places

We asked patients with either panic disorder, social phobia, or major depressive disorder and healthy control participants to describe their most frightening experience and to describe an emotionally neutral experience. Both fear and neutral autobiographical memories were audiotaped and processed through a low-pass filter that eliminated frequencies above 400 Hz, thereby abolishing semantic content but leaving paralinguistic aspects like rate, pitch, and loudness intact, and these convey emotional cues. Raters blind to content and diagnosis rated the content-filtered speech clips on emotional dimensions. The results revealed that content-filtered fear memories received significantly higher ratings on anxious, aroused, and dominant (but not sad or negative) scales than did content-filtered neutral memories, irrespective of the diagnostic status of the speaker. Content-filtered speech appears promising as an on-line probe of emotional processing during accessing of autobiographical memories.

We applied a variant of Jacoby's (1991) process dissociation procedure to parse the relative contributions of automatic and controlled processes to word-stem completion performance involving threatening, positive, and neutral material in patients with panic disorder and healthy control participants. Contrary to prediction, processing of threatening (relative to nonthreatening) information in panic disorder was not disproportionately influenced by automatic processing. We found limited evidence, however, that panic patients exhibit a baseline bias for completing stems relevant to threat relative to nonthreat stems, perhaps indicating a proneness to engage in self-generated priming of threat material.

The purpose of this study was to test a novel pictorial emotional Stroop paradigm that required participants to name the colors of filtered images on a computer screen. High (n = 20) and low (n = 20) snake-fearful participants color-named filters covering images of snakes (threat), cows (neutral), bunnies (positive), and blank screens. Each image appeared as if viewed through tinted sunglasses. The results revealed that both groups took longer to name the colors of filters covering bunnies as well as snakes relative to filters covering cows. Intensely snake-fearful individuals (n = 5), however, exhibited additional interference for snake pictures beyond that evoked by bunny pictures. Thus, pictorial cues having positive as well as negative emotional valence disproportionately captured attention. This paradigm shows promise as a nonlexical, ecologically valid approach to evaluating selective processing of emotional cues.

Patients with panic disorder are characterized by elevated scores on the Anxiety Sensitivity Index (ASI; Reiss, Peterson, Gursky, & McNally, 1986) and by interpretive, attentional, and memory biases favoring the processing of threatening information (McNally, 1994, pp. 123–132). Although longitudinal research indicates that elevated ASI scores predict the occurrence of spontaneous panic (Schmidt, Lerew, & Jackson, 1997), it is unclear whether high anxiety sensitivity is premorbidly associated with information-processing biases that may themselves reflect cognitive risk for panic. To address this issue, we had 63 individuals who reported no history of spontaneous panic perform interpretive, attentional, and memory tasks similar to those revealing threat-related cognitive biases in panic disorder patients. Subjects completed the ASI as well as other potential predictors of cognitive bias (e.g., measures of state and trait anxiety). The results indicated that ASI scores predicted one measure of interpretive bias, but little else. Although elevated ASI scores appear to reflect cognitive risk for panic attacks, cognitive bias variables rarely correlate strongly with this risk factor.

We used a directed-forgetting paradigm to investigate whether panic disorder patients cognitively avoid threatening information. To determine if hemispheric laterality predicts processing biases in this paradigm, we used dichotic listening methods to ascertain participants' auditory perceptual asymmetry (PA). Panic disorder patients and healthy control participants viewed a series of intermixed threat, positive, and neutral words, each followed by an instruction to either remember the word or forget it. They then performed free recall and recognition tests for all words, irrespective of initial instructions. Directed-forgetting effects occurred equally for all word types: both groups recalled remember-words better than forget-words. Because this task is strongly affected by encoding style, panic patients as a group do not seem to avoid encoding threat cues. However, PA analyses revealed that cognitive avoidance of threat forget-words was significantly associated with greater left hemisphere bias in the control group and nonsignificantly associated with lesser left hemisphere bias in the panic disorder group.

This study investigates lifetime prevalence rates, demographic characteristics, childhood conduct disorder and adult antisocial features, suicide attempts, and cognitive impairment in individuals with obsessive-compulsive disorder (OCD) uncomplicated by or comorbid with any other psychiatric disorder. The data are from the NIMH Epidemiological Catchment Area (ECA) study, and the current analyses compared subjects with uncomplicated OCD (no history of any other lifetime psychiatric disorder) comorbid OCD (with any other lifetime disorder), other lifetime psychiatric disorders, and no lifetime psychiatric disorders across these variables. OCD in its uncomplicated and comorbid form had significantly higher rates of childhood conduct symptoms, adult antisocial personality disorder problems, and of suicide attempts than did no or other disorders. Comorbid OCD subjects had higher rates of mild cognitive impairment on the Mini-Mental Status Exam than did subjects with other disorders. These findings suggest that a subgroup of OCD patients may have impulsive features, including childhood conduct disorder symptoms and an increased rate of suicide attempts; wider clinical attention to these outcomes is needed.

Patients with panic disorder and psychiatrically healthy control subjects performed a dual priming task whereby they viewed either lexical or non-lexical prime pairs before naming a target that had either threatening (e.g. collapse) or positive (e.g. cheerful) meaning. Lexical prime pairs comprised a threat word and a positive word, and non-lexical prime pairs comprised two rows of asterisks. Suggestive of a bias for encoding threat cues, panic disorder patients (under some conditions) were faster to name lexically primed threat targets than lexically primed positive targets. These data are consistent with the hypothesis that panic disorder is linked to an encoding bias for threatening relative positive information. A cognitive bias for selectively encoding threat cues may figure in the maintenance of anxiety states, such as panic disorder.

This study investigated lifetime prevalence rates, demographic characteristics, childhood conduct disorder and adult antisocial features, suicide attempts, and cognitive impairment in individuals with obsessive-compulsive disorder (OCD) uncomplicated by or comorbid with any other psychiatric disorder. The data are from the NIMH Epidemiologic Catchment Area (ECA) study, and the current analyses compared subjects with uncomplicated OCD (no history of any other lifetime psychiatric disorder), comorbid OCD (with any other lifetime disorder), other lifetime psychiatric disorders, and no lifetime psychiatric disorders across these variables. OCD in its uncomplicated and comorbid form had significantly higher rates of childhood conduct symptoms, adult antisocial personality disorder problems, and of suicide attempts than did no or other disorders. Comorbid OCD subjects had higher rates of mild cognitive impairment on the Mini-Mental Status Exam than did subjects with other disorders. These findings suggest that a subgroup of OCD patients may have impulsive features, including childhood conduct disorder symptoms and an increased rate of suicide attempts; wider clinical attention to these outcomes is needed.

Analysing data from the Epidemiologic Catchment Area (ECA) study, Weissman and colleagues reported that panic disorder was strongly associated with suicide attempt. However, they did not control optimally for comorbid disorders known to increase suicide risk. Reanalysing the ECA data, we controlled for comorbid disorders in the aggregate rather than one at a time when we estimated the association between panic disorder and suicide attempt. Panic disorder was not associated with an increased risk of suicide attempt. Comorbid conditions strongly influence whether people with panic disorder are at especial risk of suicide attempt.