Academic journal article Canadian Journal of Experimental Psychology

Recognition of Temporal Patterns: From Engineering to Psychology and Back Again

Academic journal article Canadian Journal of Experimental Psychology

Recognition of Temporal Patterns: From Engineering to Psychology and Back Again

Article excerpt


An overview is presented on research conducted in our lab to quantify the underlying principles behind the recognition of temporal patterns. We have been developing a theory based upon pattern matching and time-series analysis, which allows us to model and understand how humans recognize familiar patterns evolving over time and how performance degrades with noise. While our studies are primarily scientific in nature, the work has application beyond the elucidation of psychological and physiological mechanisms. We illustrate an application of these ideas to computer-based human identification through gait analysis. This study also illustrates a novel approach to interdisciplinary research by integrating experimental psychology with that of engineering design.

Résumé Cet article présente un aperçu des recherches effectuées dans notre laboratoire en vue de quantifier les principes fondamentaux de la reconnaissance des tendances temporelles. Nous avons élaboré une théorie fondée sur l'appariement des tendances et l'analyse des séries temporelles, ce qui nous permet de modéliser et de comprendre comment les humains reconnaissent les tendances familières qui évoluent avec le temps et comment le rendement diminue en raison du bruit. Bien que nos études soient surtout de nature scientifique, le travail peut être appliqué au-delà de l'éclaircissement des mécanismes psychologique et physiologique. Nous illustrons une application de ces idées pour l'identification génétique informatisée par l'entremise de l'analyse de la démarche. En intégrant la psychologie expérimentale à celle de la conception technique, cette étude jette aussi un nouvel éclairage sur la recherche multidisciplinaire.

Humans are very adept at recognizing familiar patterns. The sounds of footsteps can often reveal a person's identity; a newborn child learns early to recognize his or her parent's face; the sound of a familiar melody can be easily identified even when it is transposed or distorted. All of these examples highlight our ability to distinguish, categorize, and recall complex temporal patterns.

Our interest is in developing a systems-level theory of temporal pattern recognition that has application across the different sensory modalities. Inspired from engineering, a schematic flowchart illustrating our idea is shown in Figure 1. This flowchart illustrates the processing of information during the task of pattern classification. External stimuli that fall within an integration period are clustered into a single "processing frame." A preprocessing stage processes each frame individually to extract only that information which is necessary for pattern classification. The preprocessed frames are now matched to known templates stored either cognitively (e.g., in memory) or are hardwired in the neural configuration (e.g., preferential matching by neurons). Time-series analysis quantifies the likelihood that the sequence of images belongs to a particular pattern class. Finally, this likelihood is used as a basis for pattern classification. We provide a detailed explanation of the proposed system architecture in the next section.

Theoretical Background

In a seminal paper in the 1970s, the Swedish psychologist Gunnar Johansson demonstrated that humans are particularly sensitive to motion of a biological origin (Johansson, 1973)· By illuminating only the joints of the body, Johansson showed that the correlated motion of the joints alone was sufficient for an observer to perceive what the person was doing (e.g., Brownlow, Dixon, Egbert, & Radcliffe, 1997), to identify the gender of the person (e.g., Mather & Murdoch, 1994) or to even recognize a friend (e.g., Cutting & Kozlowski, 1977; Loula, Prasad, Harber, & Shiffrar, 2005). An example of a "Johansson display" is shown in Figure 2.

The discovery by Johansson illustrates an important example of Stage 1 ("preprocessing") in the system architecture described in Figure 1. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed


An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.