Joseph P. Dunlop

Theoretical, Statistical, and Practical Perspectives on Pattern-based Classification Approaches to the Analysis of Functional Neuroimaging Data

The goal of pattern-based classification of functional neuroimaging data is to link individual brain activation patterns to the experimental conditions experienced during the scans. These brain-reading analyses advance functional neuroimaging on three fronts. From a technical standpoint, pattern-based classifiers overcome fatal f laws in the status quo inferential and exploratory multivariate approaches by combining pattern-based analyses with a direct link to experimental variables. In theoretical terms, the results that emerge from pattern-based classifiers can offer insight into the nature of neural representations. This shifts the emphasis in functional neuroimaging studies away from localizing brain activity toward understanding how patterns of brain activity encode information. From a practical point of view, pattern-based classifiers are already well established and understood in many areas of cognitive science. These tools are familiar to many researchers and provide a quantitatively sound and qualitatively satisfying answer to most questions addressed in functional neuroimaging studies. Here, we examine the theoretical, statistical, and practical underpinnings of pattern-based classification approaches to functional neuroimaging analyses. Pattern-based classification analyses are well positioned to become the standard approach to analyzing functional neuroimaging data.

An introduction to the good, the bad, & the ugly face recognition challenge problem

The Good, the Bad, & the Ugly Face Challenge Problem was created to encourage the development of algorithms that are robust to recognition across changes that occur in still frontal faces. The Good, the Bad, & the Ugly consists of three partitions. The Good partition contains pairs of images that are considered easy to recognize. On the Good partition, the base verification rate (VR) is 0.98 at a false accept rate (FAR) of 0.001. The Bad partition contains pairs of images of average difficulty to recognize. For the Bad partition, the VR is 0.80 at a FAR of 0.001. The Ugly partition contains pairs of images considered difficult to recognize, with a VR of 0.15 at a FAR of 0.001. The base performance is from fusing the output of three of the top performers in the FRVT 2006. The design of the Good, the Bad, & the Ugly controls for pose variation, subject aging, and subject “recognizability.” Subject recognizability is controlled by having the same number of images of each subject in every partition. This implies that the differences in performance among the partitions are result of how a face is presented in each image.

Differential reward network functional connectivity in cannabis dependent and non-dependent users

BACKGROUND Emergent studies show that similar to other substances of abuse, cue-reactivity to cannabis is also associated with neural response in the brains reward pathway (Filbey et al., 2009). However, the inter-relatedness of brain regions during cue-reactivity in cannabis users remains unknown. METHODS In this study, we conducted a series of investigations to determine functional connectivity during cue-reactivity in 71 cannabis users. First, we used psychophysiological interaction (PPI) analysis to examine coherent neural response to cannabis cues. Second, we evaluated whether these patterns of network functional connectivity differentiated dependent and non-dependent users. Finally, as an exploratory analysis, we determined the directionality of these connections via Granger connectivity analyses. RESULTS PPI analyses showed reward network functional connectivity with the nucleus accumbens (NAc) seed region during cue exposure. Between-group contrasts found differential effects of dependence status. Dependent users (N=31) had greater functional connectivity with amygdala and anterior cingulate gyrus (ACG) seeds while the non-dependent users (N=24) had greater functional connectivity with the NAc, orbitofrontal cortex (OFC) and hippocampus seeds. Granger analyses showed that hippocampal and ACG activation preceded neural response in reward areas. CONCLUSIONS Both PPI and Granger analyses demonstrated strong functional coherence in reward regions during exposure to cannabis cues in current cannabis users. Functional connectivity (but not regional activation) in the reward network differentiated dependent from non-dependent cannabis users. Our findings suggest that repeated cannabis exposure causes observable changes in functional connectivity in the reward network and should be considered in intervention strategies.

Neural Effects of Positive and Negative Incentives during Marijuana Withdrawal

In spite of evidence suggesting two possible mechanisms related to drug-seeking behavior, namely reward-seeking and harm avoidance, much of the addiction literature has focused largely on positive incentivization mechanisms associated with addiction. In this study, we examined the contributing neural mechanisms of avoidance of an aversive state to drug-seeking behavior during marijuana withdrawal. To that end, marijuana users were scanned while performing the monetary incentive delay task in order to assess positive and negative incentive processes. The results showed a group x incentive interaction, such that marijuana users had greater response in areas that underlie reward processes during positive incentives while controls showed greater response in the same areas, but to negative incentives. Furthermore, a negative correlation between withdrawal symptoms and response in the amygdala during negative incentives was found in the marijuana users. These findings suggest that although marijuana users have greater reward sensitivity and less harm avoidance than controls, that attenuated amygdala response, an area that underlies fear and avoidance, was present in marijuana users with greater marijuana withdrawal symptoms. This is concordant with models of drug addiction that involve multiple sources of reinforcement in substance use disorders, and suggests the importance of strategies that focus on respective mechanisms.

The hyper-sentient addict: An exteroception model of addiction

Abstract Background: Exteroception involves processes related to the perception of environmental stimuli important for an organism’s ability to adapt to its environment. As such, exteroception plays a critical role in conditioned response. In addiction, behavioral and neuroimaging studies show that the conditioned response to drug-related cues is often associated with alterations in brain regions including the precuneus/posterior cingulate cortex, an important node within the default mode network dedicated to processes such as self-monitoring. Objective: This review aimed to summarize the growing, but largely fragmented, literature that supports a central role of exteroceptive processes in addiction. Methods: We performed a systematic review of empirical research via PubMed and Google Scholar with keywords including ‘addiction’, ‘exteroception’, ‘precuneus’, and ‘self-awareness’, to identify human behavioral and neuroimaging studies that report mechanisms of self-awareness in healthy populations, and altered self-awareness processes, specifically exteroception, in addicted populations. Results: Results demonstrate that exteroceptive processes play a critical role in conditioned cue response in addiction and serve as targets for interventions such as mindfulness training. Further, a hub of the default mode network, namely, the precuneus, is (i) consistently implicated in exteroceptive processes, and (ii) widely demonstrated to have increased activation and connectivity in addicted populations. Conclusion: Heightened exteroceptive processes may underlie cue-elicited craving, which in turn may lead to the maintenance and worsening of substance use disorders. An exteroception model of addiction provides a testable framework from which novel targets for interventions can be identified.

fMRI study of neural sensitization to hedonic stimuli in long‐term, daily cannabis users

Although there is emergent evidence illustrating neural sensitivity to cannabis cues in cannabis users, the specificity of this effect to cannabis cues as opposed to a generalized hyper‐sensitivity to hedonic stimuli has not yet been directly tested. Using fMRI, we presented 53 daily, long‐term cannabis users and 68 non‐using controls visual and tactile cues for cannabis, a natural reward, and, a sensory‐perceptual control object to evaluate brain response to hedonic stimuli in cannabis users. The results showed an interaction between group and reward type such that the users had greater response during cannabis cues relative to natural reward cues (i.e., fruit) in the orbitofrontal cortex, striatum, anterior cingulate gyrus, and ventral tegmental area compared to non‐users (cluster‐threshold z = 2.3, P < 0.05). In the users, there were positive brain‐behavior correlations between neural response to cannabis cues in fronto‐striatal‐temporal regions and subjective craving, marijuana‐related problems, withdrawal symptoms, and levels of THC metabolites (cluster‐threshold z = 2.3, P < 0.05). These findings demonstrate hyper‐responsivity, and, specificity of brain response to cannabis cues in long‐term cannabis users that are above that of response to natural reward cues. These observations are concordant with incentive sensitization models suggesting sensitization of mesocorticolimbic regions and disruption of natural reward processes following drug use. Although the cross‐sectional nature of this study does not provide information on causality, the positive correlations between neural response and indicators of cannabis use (i.e., THC levels) suggest that alterations in the reward system are, in part, related to cannabis use. Hum Brain Mapp 37:3431–3443, 2016.

Multiple Subject Barycentric Discriminant Analysis (MUSUBADA): How to Assign Scans to Categories without Using Spatial Normalization

We present a new discriminant analysis (DA) method called Multiple Subject Barycentric Discriminant Analysis (MUSUBADA) suited for analyzing fMRI data because it handles datasets with multiple participants that each provides different number of variables (i.e., voxels) that are themselves grouped into regions of interest (ROIs). Like DA, MUSUBADA (1) assigns observations to predefined categories, (2) gives factorial maps displaying observations and categories, and (3) optimally assigns observations to categories. MUSUBADA handles cases with more variables than observations and can project portions of the data table (e.g., subtables, which can represent participants or ROIs) on the factorial maps. Therefore MUSUBADA can analyze datasets with different voxel numbers per participant and, so does not require spatial normalization. MUSUBADA statistical inferences are implemented with cross-validation techniques (e.g., jackknife and bootstrap), its performance is evaluated with confusion matrices (for fixed and random models) and represented with prediction, tolerance, and confidence intervals. We present an example where we predict the image categories (houses, shoes, chairs, and human, monkey, dog, faces,) of images watched by participants whose brains were scanned. This example corresponds to a DA question in which the data table is made of subtables (one per subject) and with more variables than observations.

Effect of Age on Variability in the Production of Text-Based Global Inferences

As we age, our differences in cognitive skills become more visible, an effect especially true for memory and problem solving skills (i.e., fluid intelligence). However, by contrast with fluid intelligence, few studies have examined variability in measures that rely on one’s world knowledge (i.e., crystallized intelligence). The current study investigated whether age increased the variability in text based global inference generation–a measure of crystallized intelligence. Global inference generation requires the integration of textual information and world knowledge and can be expressed as a gist or lesson. Variability in generating two global inferences for a single text was examined in young-old (62 to 69 years), middle-old (70 to 76 years) and old-old (77 to 94 years) adults. The older two groups showed greater variability, with the middle elderly group being most variable. These findings suggest that variability may be a characteristic of both fluid and crystallized intelligence in aging.

Recognizing people from dynamic and static faces and bodies: Dissecting identity with a fusion approach

The goal of this study was to evaluate human accuracy at identifying people from static and dynamic presentations of faces and bodies. Participants matched identity in pairs of videos depicting people in motion (walking or conversing) and in ‘‘best’’ static images extracted from the videos. The type of information presented to observers was varied to include the face and body, the face-only, and the body-only. Identification performance was best when people viewed the face and body in motion. There was an advantage for dynamic over static stimuli, but only for conditions that included the body. Control experiments with multiple-static images indicated that some of the motion advantages we obtained were due to seeing multiple images of the person, rather than to the motion, per se. To computationally assess the contribution of different types of information for identification, we fused the identity judgments from observers in different conditions using a statistical learning algorithm trained to optimize identification accuracy. This fusion achieved perfect performance. The condition weights that resulted suggest that static displays encourage reliance on the face for recognition, whereas dynamic displays seem to direct attention more equitably across the body and face.

Imaging Genetics with Partial Least Squares for Mixed-Data Types (MiMoPLS)

“Imaging genetics” studies the genetic contributions to brain structure and function by finding correspondence between genetic data—such as single nucleotide polymorphisms (SNPs)—and neuroimaging data—such as diffusion tensor imaging (DTI). However, genetic and neuroimaging data are heterogenous data types, where neuroimaging data are quantitative and genetic data are (usually) categorical. So far, methods used in imaging genetics treat all data as quantitative, and this sometimes requires unrealistic assumptions about the nature of genetic data. In this article we present a new formulation of Partial Least Squares Correlation (PLSC)—called Mixed-modality Partial Least Squares (MiMoPLS)—specifically tailored for heterogeneous (mixed-) data types. MiMoPLS integrates features of PLSC and Correspondence Analysis (CA) by using special properties of quantitative data and Multiple Correspondence Analysis (MCA). We illustrate MiMoPLS with an example data set from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) with DTI and SNPs.