Back to overview

Development of the mechanisms underlying feature-based selective perception in children and young adults

Applicant Krummenacher Joseph
Number 125205
Funding scheme Project funding (Div. I-III)
Research institution Département de Psychologie Université de Fribourg
Institution of higher education University of Fribourg - FR
Main discipline Psychology
Start/End 01.06.2009 - 31.05.2011
Approved amount 38'930.00
Show all

Keywords (14)

Visual Perception; Selective Mechanisms; Cognitive Development; Learning; Memory; Mechanisms of perception; Attentional selection; Bottom-up and top-down control; Developmental psychophysics; Oculo-motor control; selective attention; dimension-based processing; development of perception; development of feature-based attention

Lay Summary (English)

Lay summary
A key cognitive ability is the selection of information relevant for the control of behavior (attention). Selection is investigated in 'visual search' experiments. Observers indicate the presence of a target object shown, in half of the trials, in an array of distractors. Targets differing from distractors by salient features (a red apple among green apples) are detected rapidly. Targets defined by feature combinations (a red small apple among red large and green large or small apples) are hard to find. Dimension-based models such as 'Guided Search' (Wolfe, 1994) assume that selection occurs in two stages. In stage one, basic features (lines, colors, etc.) are extracted and represented in feature-based modules. In stage two, dimension-based signals are generated coding the relative saliency of display locations. The more features differ from neighboring features, the higher their saliency.Until recently, it was generally accepted that saliency generation is an automatic and stimulus-driven processes and that saliency cannot be modulated by top-down signals. However, research shows that dimension-based saliency signals need to be weighted (amplified) to be able to guide attention and that dimensional weight is top-down modulable. Dimension-based weight patterns persist into the next trials and affect search performance systematically: Repeating the target dimension across trials expedites responses; changing dimensions slows reactions. Thus, dimension-based inter-trial effects constitute markers of dimension-based processing. Objects are characterized by specific features (red, small) rather then dimensions (color, size), thus, it is surprising that inter-trial modulations are dimension-, rather than feature-, based. The present study aims to investigate the functional role and development of dimension-based effects in visual processing in children and young adults. Results of a pilot feature search study with 100 participants aged between six and 20 years suggest that dimension-based effects develop at age eight. Moreover, a decrease in overall reaction times can be attributed to both motor and cognitive development.Four experimental series are proposed to assess the development of dimension-based processing in singleton feature, conjunction, and compound search tasks. Cuing is used to assess top-down effects. Further, the relationship between oculomotor processes and attention is investigated. As little is known about the latencies of saccadic eye movements in children, the research project is expected to contribute both to basic research and to issues in developmental psychology.
Direct link to Lay Summary Last update: 21.02.2013

Responsible applicant and co-applicants


Associated projects

Number Title Start Funding scheme
133888 Selective attention and eye movements: An integrative cognitive neuropsychological approach 01.12.2010 R'EQUIP
114415 Theoretical and empirical integration of process- and parameter-based accounts of visual perception 01.10.2006 ProDoc
110543 Selective visual processing: An integrative neurocognitive account based on psychological research methods 01.09.2006 SNSF Professorships
114405 Mechanismen der menschlichen Wahrnehmung: Selektion, Integration, Kontrolle, Lernen und Gedächtnis 01.10.2006 ProDoc


Among the core cognitive abilities in humans is the selection of information relevant for the control of current thinking and behavior and/or the concurrent de-selection of irrelevant stimulation. The mechanisms underlying selection are referred to as processes of ‘selective attention’ or ‘attentional processes’. Current influential theoretical models of visual selective processing assume two processing stages: on a first stage, basic visual information is extracted from a scene and represented in a set of independent feature- or dimension-based modules. It is assumed that these representations are used to compute a type of activity that signals locations in a visual scene that constitute areas containing potentially interesting or important information, that is, areas where attention should be directed at. It is important to note that this so called saliency signal does not carry any information as to the characteristics of the objects present at a particular location in a visual field, rather it codes degrees of conspicuity. Saliency signals are generated for the different visual dimensions and then integrated into an overall saliency representation. It was assumed that saliency representations are generated in an automatic fashion by processes that are stimulus-driven and that saliency activity cannot be modulated by stimulus- and goal- or expectancy-driven signals. Recent (own) research has shown, however, that dimension-based saliency signals need to be weighted (amplified) in order to exceed a threshold constituting a lower bound for activities that are able to guide attention to display locations. Interestingly, the dimension-based weight pattern established in a given experimental trial persists into the next trial to affect search performance in systematic fashion: If the dimension defining the target of the preceding trial is repeated in the current trial, reaction times are expedited, if the dimension changes, reaction times are slowed. As the exact functional role of these dimension-based reaction time costs and benefits remains unclear, in a collaboration between two research modules of the SNSF Pro*Doc doctoral school “Processes of Human Perception” we decided to investigate potential causes of the effects in a developmental approach. In a visual search experiment, more than 100 children and young adults aged between six and 20 years were instructed to detect targets differing from distractors by color or orientation. Reaction times were analyzed dependent on whether the target-defining dimension changed across consecutive trials or whether it remained the same. Results of this study suggest that dimension-based effects develop at an age of about eight years; children younger than eight show feature- rather than dimension-based effects. In order to investigate in more detail the exact nature of the mechanisms underlying dimension- (or category-) and feature-based effects in visual processing as well as their development we propose to extend the present approach by running a series of additional experiments with children and young adults between six and 20 years of age. Since the payment of experimenters conducting the empirical investigations (paid until now by the department of cognitive and developmental psychology) exceeds the available financial means we apply for funding to extend the current project.In detail, four experimental series are proposed, that will not only extend the current knowledge on category- or dimension-based processing and its development in children but also provide valuable results in basic research. As an example, very little is known about the latencies of eye movements in children, an issue we propose to address in the second of two research foci; a first focus is dedicated to the collection of behavioral data using different types of search tasks including feature, conjunction and compound search.The planned studies will be carried out in cooperation by Marcello Indino from the University of Zurich and Anna Grubert from the University of Fribourg. They are both members of the SNF Pro*Doc doctoral school “Processes of Human Perception” and have been collaborating very successfully in the first study reported in the present proposal.