Item

Abstract

Studies in the field of visual object recognition generally report observed human performance with 2D still images e.g. photographs, line-drawings. One of the main reasons for doing so
stems from the ready availability of such stimuli for experimentation (for example, see
http://www.cog.brown.edu/~tarr/projects/databank.html). Human visual perception, however,
is a dynamic process - as the result of either an active observer or a moving target, the visual
experience is rarely static. Hence, it is important to question whether such findings
realistically portray daily human behavior. Recent experiments using dynamic stimuli have
shown that human performance can differ as a result of introducing natural motion
information to the studied object; for example, there is a recognition benefit for when faces
are seen moving (e.g., Toole et al, 2002). Such evidence clearly suggests that object motion
plays a non-trivial role in visual recognition. Nonetheless, there are challenges - both
technical and experimental - that a researcher ought to consider when using dynamic stimuli.
Here, I will discuss some of these issues as well as the steps that were adopted, in my
research, to overcome them. In particular, I will describe how different types of dynamic
stimuli could be generated for various experiments in novel object and face learning, as well
as some of software and hardware available for this undertaking. In addition, I will briefly
discuss how such stimuli could be presented in psychophysical experiments, such as to
control for possible artifacts e.g., timing errors.