Welcome!

GitHubNo Facebook

Nicholas M. Van Horn, Ph.D.
Associate Professor
Department of Psychology
Capital University
Email: nvanhorn [at] protonmail d o t com
PGP Public Fingerprint: FAD0 7D90 1D28 748F 431C ACC6 C7FF 6249 02D5 E1E8
Nicholas Van Horn Public Key

Research Interests

I am a cognitive psychologist with a focus in visual psychophysics and computational and mathematical modeling of human visual processing. My colleagues and I use these approaches to investigate the neural mechanisms responsible for memory and learning, recognition of visual-spatial object relations, and biological psychology broadly construed. In addition, I also develop and promote software and technical solutions geared toward students and faculty interested in research, including student-friendly statistical tools, social media data acquisition and analysis, data visualization, and workflows related to tasks in the social sciences.

Visual Cognition

Visual Working Memory

Visual short-term memory (VSTM) is the mechanism by which task-relevant perceptual information is selectively chosen and maintained across relatively short time periods. Although short-lived, this information is critical for completing many visual tasks both inside and outside the laboratory, and provides continuity across our visual experience, particularly in the face of frequent eye movements. In a recent series of studies, my collaborators and I have been examining interactions between the contents of VSTM and sensory input. Previously thought to be shielded from the influence of short-term memory, our work has demonstrated that the contents of working memory can bias the ways in which we perceive the world around us.

Mechanisms of Visual Perceptual Learning

Research has established that training on perceptual tasks can lead to improvements in early sensory areas of the brain previously thought to be fixed after adolescence. This kind of learning is characterized by slow improvements that are largely specific to properties of the stimuli and/or task, in stark contrast to the quicker, more generalizable learning typically associated with cognitive improvements. I am interested in the neural mechanisms driving this form of learning. A prominent explanation of this mechanism is that the sensory representations involved in encoding visual input are somehow changed by practice. However, my recent research supports the Selective Reweighting Model of perceptual learning which posits that learning occurs at the interface between perception and decision making areas of the brain. See my research for more information.

Models of Visual Object Recognition

Despite the increasing power of modern computers, human observers (even children) remain vastly superior to the best computer vision algorithms at quickly and accurately identifying objects in the visual scene. Study in the field of human object recognition is often informed by work in computer vision and engineering, and vice versa. This cooperation has led to tremendous advances in the development of models of object recognition in recent years. However, a common strategy of such models is to train a classifier on an unordered collection of “features” extracted from an image corpus. Although some positional informational is retained through redundancy in the features, my research has helped to identify a critical shortcoming in this family of models: relational information (the relative position of features with respect to one another) is almost completely lost in the models, yet is used as a primary source of information by human observers.

Data Science & Social Media Analysis

Aside from my research in visual cognition, I am intersted in Data Science broadly construed. I have 15+ years of programming experience and 6+ years of extensive experience in mathematical and computational modeling of large data sets, primarily using R, Matlab/Octave, and Python. In addition to my ongoing research in vision, I am currently involved in a collaboration with researchers at the University of Florida and Clemson University on the analysis of social network and news media outlets. I have written custom software in R for this project to facilitate automated mining and analysis of data on the web. Check out my research for more information on this and more.

Software

My research and hobbies have led me to write many software packages and tools for managing my work-flow. I am an ardent supporter and contributor to free and open source software. My primary languages for research are R, Matlab/Octave, and Python, but I have written many tools and libraries in Emacs Lisp, Scheme (Chicken), and Clojure. The results of many of these projects will be shared on this site and/or my Github page.

Selected Scholarly Work

For a complete list of my publications and conference activities, see my curriculum vitae.

Van Horn, Nicholas M. and Petrov, Alexander A. (2013). Cross-Talk Between Visual Short-Term Memory and Low-Level Vision: Evidence for Interactions Across Shared Neural Resources. Proceedings of the 2013 Annual Meeting of the Vision Sciences Society.

Petrov, Alexander and Van Horn, Nicholas M. (2012). Motion aftereffect duration is not changed by perceptual learning: Evidence against the representation modification hypothesis. Vision Research, 61, 4-14.

Petrov, Alexander A. and Van Horn, Nicholas M. and Todd, James (2011). The visual identification of relational categories. Journal of Vision, 11(12), 1-11.

Petrov, Alexander A. and Van Horn, Nicholas M. and Ratcliff, Roger (2011). Dissociable perceptual learning mechanisms revealed by diffusion-model analysis of the patterns of specificity. Psychonomic Bulletin & Review, 18(3), 490-497.