Refine
Year of publication
Document Type
- Article (99)
- Postprint (10)
- Other (5)
- Conference Proceeding (3)
- Preprint (2)
Keywords
- Eye movements (10)
- eye movements (8)
- scene viewing (6)
- Reading (5)
- attention (5)
- saccades (4)
- Bayesian inference (3)
- Computational modelling (3)
- dynamical model (3)
- eye-movement control (3)
Institute
- Department Psychologie (119) (remove)
In real-world scene perception, human observers generate sequences of fixations to move image patches into the high-acuity center of the visual field. Models of visual attention developed over the last 25 years aim to predict two-dimensional probabilities of gaze positions for a given image via saliency maps. Recently, progress has been made on models for the generation of scan paths under the constraints of saliency as well as attentional and oculomotor restrictions. Experimental research demonstrated that task constraints can have a strong impact on viewing behavior. Here, we propose a scan-path model for both fixation positions and fixation durations, which include influences of task instructions and interindividual differences. Based on an eye-movement experiment with four different task conditions, we estimated model parameters for each individual observer and task condition using a fully Bayesian dynamical modeling framework using a joint spatial-temporal likelihood approach with sequential estimation. Resulting parameter values demonstrate that model properties such as the attentional span are adjusted to task requirements. Posterior predictive checks indicate that our dynamical model can reproduce task differences in scan-path statistics across individual observers.
Lisa Schwetlick et al. present a computational model linking visual scan path generation in scene viewing to physiological and experimental work on perisaccadic covert attention, the act of attending to an object visually without obviously moving the eyes toward it. They find that integrating covert attention into predictive models of visual scan paths greatly improves the model's agreement with experimental data. <br /> How we perceive a visual scene depends critically on the selection of gaze positions. For this selection process, visual attention is known to play a key role in two ways. First, image-features attract visual attention, a fact that is captured well by time-independent fixation models. Second, millisecond-level attentional dynamics around the time of saccade drives our gaze from one position to the next. These two related research areas on attention are typically perceived as separate, both theoretically and experimentally. Here we link the two research areas by demonstrating that perisaccadic attentional dynamics improve predictions on scan path statistics. In a mathematical model, we integrated perisaccadic covert attention with dynamic scan path generation. Our model reproduces saccade amplitude distributions, angular statistics, intersaccadic turning angles, and their impact on fixation durations as well as inter-individual differences using Bayesian inference. Therefore, our result lend support to the relevance of perisaccadic attention to gaze statistics.
Bottom-up and top-down as well as low-level and high-level factors influence where we fixate when viewing natural scenes. However, the importance of each of these factors and how they interact remains a matter of debate. Here, we disentangle these factors by analyzing their influence over time. For this purpose, we develop a saliency model that is based on the internal representation of a recent early spatial vision model to measure the low-level, bottom-up factor. To measure the influence of high-level, bottom-up features, we use a recent deep neural network-based saliency model. To account for top-down influences, we evaluate the models on two large data sets with different tasks: first, a memorization task and, second, a search task. Our results lend support to a separation of visual scene exploration into three phases: the first saccade, an initial guided exploration characterized by a gradual broadening of the fixation density, and a steady state that is reached after roughly 10 fixations. Saccade-target selection during the initial exploration and in the steady state is related to similar areas of interest, which are better predicted when including high-level features. In the search data set, fixation locations are determined predominantly by top-down processes. In contrast, the first fixation follows a different fixation density and contains a strong central fixation bias. Nonetheless, first fixations are guided strongly by image properties, and as early as 200 ms after image onset, fixations are better predicted by high-level information. We conclude that any low-level, bottom-up factors are mainly limited to the generation of the first saccade. All saccades are better explained when high-level features are considered, and later, this high-level, bottom-up control can be overruled by top-down influences.
Bottom-up saliency is often cited as a factor driving the choice of fixation locations of human observers, based on the (partial) success of saliency models to predict fixation densities in free viewing. However, these observations are only weak evidence for a causal role of bottom-up saliency in natural viewing behaviour. To test bottom-up saliency more directly, we analyse the performance of a number of saliency models---including our own saliency model based on our recently published model of early visual processing (Schütt & Wichmann, 2017, JoV)---as well as the theoretical limits for predictions over time. On free viewing data our model performs better than classical bottom-up saliency models, but worse than the current deep learning based saliency models incorporating higher-level information like knowledge about objects. However, on search data all saliency models perform worse than the optimal image independent prediction. We observe that the fixation density in free viewing is not stationary over time, but changes over the course of a trial. It starts with a pronounced central fixation bias on the first chosen fixation, which is nonetheless influenced by image content. Starting with the 2nd to 3rd fixation, the fixation density is already well predicted by later densities, but more concentrated. From there the fixation distribution broadens until it reaches a stationary distribution around the 10th fixation. Taken together these observations argue against bottom-up saliency as a mechanistic explanation for eye movement control after the initial orienting reaction in the first one to two saccades, although we confirm the predictive value of early visual representations for fixation locations. The fixation distribution is, first, not well described by any stationary density, second, is predicted better when including object information and, third, is badly predicted by any saliency model in a search task.
Dynamical models of cognition play an increasingly important role in driving theoretical and experimental research in psychology. Therefore, parameter estimation, model analysis and comparison of dynamical models are of essential importance. In this article, we propose a maximum likelihood approach for model analysis in a fully dynamical framework that includes time-ordered experimental data. Our methods can be applied to dynamical models for the prediction of discrete behavior (e.g., movement onsets); in particular, we use a dynamical model of saccade generation in scene viewing as a case study for our approach. For this model, the likelihood function can be computed directly by numerical simulation, which enables more efficient parameter estimation including Bayesian inference to obtain reliable estimates and corresponding credible intervals. Using hierarchical models inference is even possible for individual observers. Furthermore, our likelihood approach can be used to compare different models. In our example, the dynamical framework is shown to outperform nondynamical statistical models. Additionally, the likelihood based evaluation differentiates model variants, which produced indistinguishable predictions on hitherto used statistics. Our results indicate that the likelihood approach is a promising framework for dynamical cognitive models.
Process-oriented theories of cognition must be evaluated against time-ordered observations. Here we present a representative example for data assimilation of the SWIFT model, a dynamical model of the control of fixation positions and fixation durations during natural reading of single sentences. First, we develop and test an approximate likelihood function of the model, which is a combination of a spatial, pseudo-marginal likelihood and a temporal likelihood obtained by probability density approximation Second, we implement a Bayesian approach to parameter inference using an adaptive Markov chain Monte Carlo procedure. Our results indicate that model parameters can be estimated reliably for individual subjects. We conclude that approximative Bayesian inference represents a considerable step forward for computational models of eye-movement control, where modeling of individual data on the basis of process-based dynamic models has not been possible so far.
Skilled reading requires information processing of the fixated and the not-yet-fixated words to generate precise control of gaze. Over the last 30 years, experimental research provided evidence that word processing is distributed across the perceptual span, which permits recognition of the fixated (foveal) word as well as preview of parafoveal words to the right of fixation. However, theoretical models have been unable to differentiate the specific influences of foveal and parafoveal information on saccade control. Here we show how parafoveal word difficulty modulates spatial and temporal control of gaze in a computational model to reproduce experimental results. In a fully Bayesian framework, we estimated model parameters for different models of parafoveal processing and carried out large-scale predictive simulations and model comparisons for a gaze-contingent reading experiment. We conclude that mathematical modeling of data from gaze-contingent experiments permits the precise identification of pathways from parafoveal information processing to gaze control, uncovering potential mechanisms underlying the parafoveal contribution to eye-movement control.
Skilled reading requires information processing of the fixated and the not-yet-fixated words to generate precise control of gaze. Over the last 30 years, experimental research provided evidence that word processing is distributed across the perceptual span, which permits recognition of the fixated (foveal) word as well as preview of parafoveal words to the right of fixation. However, theoretical models have been unable to differentiate the specific influences of foveal and parafoveal information on saccade control. Here we show how parafoveal word difficulty modulates spatial and temporal control of gaze in a computational model to reproduce experimental results. In a fully Bayesian framework, we estimated model parameters for different models of parafoveal processing and carried out large-scale predictive simulations and model comparisons for a gaze-contingent reading experiment. We conclude that mathematical modeling of data from gaze-contingent experiments permits the precise identification of pathways from parafoveal information processing to gaze control, uncovering potential mechanisms underlying the parafoveal contribution to eye-movement control.
Skilled reading requires information processing of the fixated and the not-yet-fixated words to generate precise control of gaze. Over the last 30 years, experimental research provided evidence that word processing is distributed across the perceptual span, which permits recognition of the fixated (foveal) word as well as preview of parafoveal words to the right of fixation. However, theoretical models have been unable to differentiate the specific influences of foveal and parafoveal information on saccade control. Here we show how parafoveal word difficulty modulates spatial and temporal control of gaze in a computational model to reproduce experimental results. In a fully Bayesian framework, we estimated model parameters for different models of parafoveal processing and carried out large-scale predictive simulations and model comparisons for a gaze-contingent reading experiment. We conclude that mathematical modeling of data from gaze-contingent experiments permits the precise identification of pathways from parafoveal information processing to gaze control, uncovering potential mechanisms underlying the parafoveal contribution to eye-movement control.