What can saliency models predict about eye movements? Spatial and sequential aspects of fixations during encoding and recognition. Foulsham, T. & Underwood, G. J Vis, 8:1--17, 2008. abstract bibtex Saliency map models account for a small but significant amount of the variance in where people fixate, but evaluating these models with natural stimuli has led to mixed results. In the present study, the eye movements of participants were recorded while they viewed color photographs of natural scenes in preparation for a memory test (encoding) and when recognizing them later. These eye movements were then compared to the predictions of a well defined saliency map model (L. Itti & C. Koch, 2000), in terms of both individual fixation locations and fixation sequences (scanpaths). The saliency model is a significantly better predictor of fixation location than random models that take into account bias toward central fixations, and this is the case at both encoding and recognition. However, similarity between scanpaths made at multiple viewings of the same stimulus suggests that repetitive scanpaths also contribute to where people look. Top-down recapitulation of scanpaths is a key prediction of scanpath theory (D. Noton & L. Stark, 1971), but it might also be explained by bottom-up guidance. The present data suggest that saliency cannot account for scanpaths and that incorporating these sequences could improve model predictions.
@article{ Foulsham_Underwood08,
author = {Foulsham, T. and Underwood, G.},
title = {{W}hat can saliency models predict about eye movements? {S}patial
and sequential aspects of fixations during encoding and recognition},
journal = {J Vis},
year = {2008},
volume = {8},
pages = {1--17},
abstract = {Saliency map models account for a small but significant amount of
the variance in where people fixate, but evaluating these models
with natural stimuli has led to mixed results. In the present study,
the eye movements of participants were recorded while they viewed
color photographs of natural scenes in preparation for a memory test
(encoding) and when recognizing them later. These eye movements were
then compared to the predictions of a well defined saliency map model
(L. Itti & C. Koch, 2000), in terms of both individual fixation locations
and fixation sequences (scanpaths). The saliency model is a significantly
better predictor of fixation location than random models that take
into account bias toward central fixations, and this is the case
at both encoding and recognition. However, similarity between scanpaths
made at multiple viewings of the same stimulus suggests that repetitive
scanpaths also contribute to where people look. Top-down recapitulation
of scanpaths is a key prediction of scanpath theory (D. Noton & L.
Stark, 1971), but it might also be explained by bottom-up guidance.
The present data suggest that saliency cannot account for scanpaths
and that incorporating these sequences could improve model predictions.}
}
Downloads: 0
{"_id":"N6LWPNbTThAEAcbtZ","bibbaseid":"foulsham-underwood-whatcansaliencymodelspredictabouteyemovementsspatialandsequentialaspectsoffixationsduringencodingandrecognition-2008","downloads":0,"creationDate":"2015-02-08T05:14:43.325Z","title":"What can saliency models predict about eye movements? Spatial and sequential aspects of fixations during encoding and recognition","author_short":["Foulsham, T.","Underwood, G."],"year":2008,"bibtype":"article","biburl":"http://cnslab.mb.jhu.edu/niebase.bib","bibdata":{"abstract":"Saliency map models account for a small but significant amount of the variance in where people fixate, but evaluating these models with natural stimuli has led to mixed results. In the present study, the eye movements of participants were recorded while they viewed color photographs of natural scenes in preparation for a memory test (encoding) and when recognizing them later. These eye movements were then compared to the predictions of a well defined saliency map model (L. Itti & C. Koch, 2000), in terms of both individual fixation locations and fixation sequences (scanpaths). The saliency model is a significantly better predictor of fixation location than random models that take into account bias toward central fixations, and this is the case at both encoding and recognition. However, similarity between scanpaths made at multiple viewings of the same stimulus suggests that repetitive scanpaths also contribute to where people look. Top-down recapitulation of scanpaths is a key prediction of scanpath theory (D. Noton & L. Stark, 1971), but it might also be explained by bottom-up guidance. The present data suggest that saliency cannot account for scanpaths and that incorporating these sequences could improve model predictions.","author":["Foulsham, T.","Underwood, G."],"author_short":["Foulsham, T.","Underwood, G."],"bibtex":"@article{ Foulsham_Underwood08,\n author = {Foulsham, T. and Underwood, G.},\n title = {{W}hat can saliency models predict about eye movements? {S}patial\n\tand sequential aspects of fixations during encoding and recognition},\n journal = {J Vis},\n year = {2008},\n volume = {8},\n pages = {1--17},\n abstract = {Saliency map models account for a small but significant amount of\n\tthe variance in where people fixate, but evaluating these models\n\twith natural stimuli has led to mixed results. In the present study,\n\tthe eye movements of participants were recorded while they viewed\n\tcolor photographs of natural scenes in preparation for a memory test\n\t(encoding) and when recognizing them later. These eye movements were\n\tthen compared to the predictions of a well defined saliency map model\n\t(L. Itti & C. Koch, 2000), in terms of both individual fixation locations\n\tand fixation sequences (scanpaths). The saliency model is a significantly\n\tbetter predictor of fixation location than random models that take\n\tinto account bias toward central fixations, and this is the case\n\tat both encoding and recognition. However, similarity between scanpaths\n\tmade at multiple viewings of the same stimulus suggests that repetitive\n\tscanpaths also contribute to where people look. Top-down recapitulation\n\tof scanpaths is a key prediction of scanpath theory (D. Noton & L.\n\tStark, 1971), but it might also be explained by bottom-up guidance.\n\tThe present data suggest that saliency cannot account for scanpaths\n\tand that incorporating these sequences could improve model predictions.}\n}","bibtype":"article","id":"Foulsham_Underwood08","journal":"J Vis","key":"Foulsham_Underwood08","pages":"1--17","title":"What can saliency models predict about eye movements? Spatial and sequential aspects of fixations during encoding and recognition","type":"article","volume":"8","year":"2008","bibbaseid":"foulsham-underwood-whatcansaliencymodelspredictabouteyemovementsspatialandsequentialaspectsoffixationsduringencodingandrecognition-2008","role":"author","urls":{},"downloads":0},"search_terms":["saliency","models","predict","eye","movements","spatial","sequential","aspects","fixations","during","encoding","recognition","foulsham","underwood"],"keywords":[],"authorIDs":[],"dataSources":["ErLXoH8mqSjESnrN5"]}