11 research outputs found
Distracted by Previous Experience: Integrating Selection History, Current Task Demands and Saliency in an Algorithmic Model- Springer2024
No full textContent
This repository contains the files and data necessary to recreate the results from the paper
N.Meibodi, H.Abbasi, A. Schuboe, D. Endres (2021) A Model of Selection History in Visual Attention, Proceedings of the 2021 Conference of the Society of Cognitive Science, Vienna, Austria.
File list
attention_models_cogsci.py: run this file in python 3.8 to recreate results:
• maps_weights.pdf: bar plots to compare the maps’ weights between groups (color and shape). Fig5 in the paper.
• evidence_par.pdf: model evidences on bar plots to compare three models. Fig4 in the paper.
• model_evidence_par.xlsx: model evidences are saved in this file, when models are fitted across participants.
• model_evidence_group.xlsx: model evidences are saved in this file, when models are fitted across groups.
• CG.pdf and SG.pdf: Ex-Gaussian distributions of reaction times, fitted on the data and also model predicted distributions, for color group and shape group participants. Fig6 in the paper. torch_hessian.py: computation of Hessian matrix, needs to be in the same directory as 'attention_models_cogsci.py'
Data_CogSci2021.xlsx: Data from which models are learned
'Description of the uploaded dataset - CogSci 2021.docx': Description of data format and content.
paper_cogsci.pdf: preprint of the proceedings paper.
Acknowledgements
This work was supported by the SFB-TRR 135 'Cardinal Mechanisms of Perception', projects C6 and B3, and 'The Adaptive Mind', funded by the Excellence Program of the Hessian Ministry for Science and the Arts.This repository contains the files and data necessary to recreate the results from the paper Meibodi, N., Abbasi, H., Schubö, A. et al. Distracted by Previous Experience: Integrating Selection History, Current Task Demands and Saliency in an Algorithmic Model. Comput Brain Behav 7, 268–285 (2024). https://doi.org/10.1007/s42113-024-00197-6
Please go to Version 2 if you are interested to see the files related to the other pubplication N.Meibodi, H.Abbasi, A. Schuboe, D. Endres (2021) A Model of Selection History in Visual Attention, Proceedings of the 2021 Conference of the Society of Cognitive Science, Vienna, Austria.DFG
Sensorimotor processes are not a source of much noise: sensorimotor and decision components of reaction times
No full textThis repository contains the files necessary to recreate the results from the paper:
N.Meibodi, A. Schuboe, D. Endres (2022) Sensorimotor processes are not a source of much noise: sensorimotor and decision components of reaction times, Proceedings of the 2022 Conference of the Society of Cognitive Science, Toronto, Canada.
# Acknowledgements
This work was supported by the SFB-TRR 135 (222641018) 'Cardinal Mechanisms of Perception', projects C6 and B3, and 'The Adaptive Mind', funded by the Excellence Program of the Hessian Ministry for Science and the Arts.DFG, HMW
Aufmerksamkeit messen: von der Reaktionszeit zu Modellen der visuellen Aufmerksamkeit
No full textModels of visual attention have been widely proposed over the last two decades. Researchers in different disciplines, such as psychology and engineering, are interested in these models in order to understand human perceptual mechanisms and/or build algorithms which mimic the attentional processes for some applications (e.g. robotics).
In this dissertation I modeled the effect of learning experiences on attentional guidance. The presented model is an algorithmic-level model which links display inputs to the participants' reaction times. This dissertation consists of three studies.
In the first study the role of selection history -as the effect of learning from the practice phase of the experiment on the main phase- is investigated. I also tested dimension-level (e.g. color and shape) and feature-level (e.g. blue and red) selection histories. The results showed the version of the model which includes selection history (on feature-level), beside stimulus-driven (bottom-up) and goal-driven (top-down) control mechanisms, is best suited for a quantitative description of the participants' reaction times.
In the second study, I investigated the importance of intertrial priming -the effect of a previous trial on the current one- as well as the importance of each feature map (color, shape or orientation) in the model predictions. It was shown that by including the effect of intertrial priming a better description of the behavioral database can be achieved. Additionally, excluding any of the feature maps deteriorates the model predictions.
In the third study, I proposed a model to decompose reaction times -into decision and sensorimotor components- as a prerequisite of RT modeling. This study will help us introduce more accurate attention models. Furthermore, it can support cognitive studies to better investigate the effect of certain factors (e.g. age and mental disorders) on motor system vs. decision making.
The proposed attention model (in the first and the second study) is one of the first models that includes the selection history effect on guiding attention. This model can capture the between-group differences where each group of participants had a different learning experience. The model considers total reaction times of each participant. But attention can influence reaction times by affecting different cognitive processes. The third study introduces a method which helps us look at each process (and its relevant reaction time component) independently
Recommended from our members
Sensorimotor processes are not a source of much noise: Sensory-motor and decision components of reaction times
No full textStatistical descriptions of reaction times are central components of quantitative attention models. It is often assumed that total reaction time is comprised of various components, e.g. sensory delays, decision making and motor execution contributions. We use machine learning to decompose observed total reaction times into sensorimotor and decision components, and evaluate which model assumptions maximize approximate Bayesian model evidence (free energy or evidence lower bound). We find that an inverse Gaussian decision time distribution combined with a very narrow Gaussian sensorimotor distribution can best explain human reaction time data. We also model outliers explicitly by a uniform background distribution. We find that the model assigns a small fraction of datapoints to this outlier distribution
Recommended from our members
A model of selection history in visual attention
No full textAttention can be biased by the previous learning and experience. We present an algorithmic-level model of this bias in visual attention that predicts quantitatively how bottom-up, top-down and selection history compete to control attention. In the model, the output of saliency maps as bottom-up guidance interacts with a history map that encodes learning effects and a top-down task control to prioritize visual features. We test the model on a reaction-time (RT) data set from the experiment presented in (Feldmann-Wustefeld, Uengoer, & Schubö, 2015). The model accurately predicts parameters of reaction time distributions from an integrated priority map that is comprised of an optimal, weighted combination of separate maps. Analysis of the weights confirms learning history effects on attention guidance
Distracted by Previous Experience: Integrating Selection History, Current Task Demands and Saliency in an Algorithmic Model
No full textAttention can be biased by previous learning and experience. We present an algorithmic-level model of this selection history bias in visual attention that predicts quantitatively how stimulus-driven processes, goal-driven control and selection history compete to control attention. In the model, the output of saliency maps as stimulus-driven guidance interacts with a history
map that encodes learning effects and a goal-driven task control to prioritize visual features. The model works on coded features rather than image pixels which is common in many traditional saliency models. We test the model on a reaction time (RT) data from a psychophysical experiment. The model accurately predicts parameters of reaction time distributions from an integrated priority map that is comprised of an optimal, weighted combination of separate maps. Analysis of the weights confirms selection history effects on attention guidance. The model is able to capture individual differences between participants’ RTs and response probabilities per group. Moreover, we demonstrate that a model with a reduced set of maps performs worse, indicating that integrating history, saliency and task information are required for a quantitative description of human attention. Besides, we show that adding intertrial effect to the model (as another lingering bias) improves the model’s predictive performance
Incidentally diagnosed multiple intradural extramedullary spinal hydatidosis in a young adult: A case report and review of the literature
No full textKey Clinical Message Although quite rare, vertebral hydatidosis should always be considered as a differential diagnosis for spinal presentations, particularly in endemic areas for echinococcosis. Abstract In this paper, we report a rare case of asymptomatic multiple intradural, extramedullary spinal hydatidosis, incidentally diagnosed in a patient with signs and symptoms of a true protruded disc. Although quite rare, vertebral hydatidosis should always be considered as a differential diagnosis for spinal presentations, particularly in endemic areas for echinococcosis
Utility of the exercise electrocardiogram testing in sudden cardiac death risk stratification
No full textBackground Sudden cardiac death (SCD) remains a major public health problem. Current established criteria identifying those at risk of sudden arrhythmic death, and likely to benefit from implantable cardioverter defibrillators (ICDs), are neither sensitive nor specific. Exercise electrocardiogram (ECG) testing was traditionally used for information concerning patients' symptoms, exercise capacity, cardiovascular function, myocardial ischemia detection, and hemodynamic responses during activity in patients with hypertrophic cardiomyopathy. Methods We conducted a systematic review of MEDLINE on the utility of exercise ECG testing in SCD risk stratification. Results Exercise testing can unmask suspected primary electrical diseases in certain patients (catecholaminergic polymorphic ventricular tachycardia or concealed long QT syndrome) and can be effectively utilized to risk stratify patients at an increased (such as early repolarization syndrome and Brugada syndrome) or decreased risk of SCD, such as the loss of preexcitation on exercise testing in asymptomatic Wolff-Parkinson-White syndrome. Conclusions Exercise ECG testing helps in SCD risk stratification in patients with and without arrhythmogenic hereditary syndromes. © 2014 Wiley Periodicals, Inc.Adler A, 2012, HEART RHYTHM, V9, P901, DOI 10.1016-j.hrthm.2012.01.026; Atta S, 2012, J CLIN EXP CARDIOLOG, V3, P223; Bastiaenen R, 2013, HEART RHYTHM, V10, P247, DOI 10.1016-j.hrthm.2012.10.032; Bershader RS BC, 2007, HEART RHYTHM, V4, pS138; Calloe K, 2013, CIRC-ARRHYTHMIA ELEC, V6, P177, DOI 10.1161-CIRCEP.112.974220; Chattha IS, 2010, HEART RHYTHM, V7, P906, DOI 10.1016-j.hrthm.2010.03.006; Cohen MI, 2012, HEART RHYTHM, V9, P1006, DOI 10.1016-j.hrthm.2012.03.050; Elhendy A, 2002, AM J CARDIOL, V90, P95, DOI 10.1016-S0002-9149(02)02428-1; Engel G, 2004, CURR PROB CARDIOLOGY, V29, P365, DOI 10.1016-j.cpcardiol.2004.02.007; Frolkis JP, 2003, NEW ENGL J MED, V348, P781, DOI 10.1056-NEJMoa022353; Gimeno JR, 2009, EUR HEART J, V30, P2599, DOI 10.1093-eurheartj-ehp327; Goldberger Jeffrey J, 2008, Heart Rhythm, V5, pe1, DOI 10.1016-j.hrthm.2008.05.031; Haissaguerre M, 2008, NEW ENGL J MED, V358, P2016, DOI 10.1056-NEJMoa071968; HINDMAN MC, 1973, ANN INTERN MED, V79, P654; Horner JM, 2011, HEART RHYTHM, V8, P1698, DOI 10.1016-j.hrthm.2011.05.018; Josephson ME, 2000, ANN INTERN MED, V133, P901; Kentta T, 2012, HEART RHYTHM, V9, P1083, DOI 10.1016-j.hrthm.2012.02.030; Lahat H, 2001, AM J HUM GENET, V69, P1378, DOI 10.1086-324565; Laitinen PJ, 2001, CIRCULATION, V103, P485; Makimoto H, 2010, J AM COLL CARDIOL, V56, P1576, DOI 10.1016-j.jacc.2010.06.033; MARIEB MA, 1990, AM J CARDIOL, V66, P172, DOI 10.1016-0002-9149(90)90583-M; Meli AC, 2011, CIRC RES, V109, P281, DOI 10.1161-CIRCRESAHA.111.244970; Morshedi-Meibodi A, 2004, CIRCULATION, V109, P2417, DOI 10.1161-01.CIR.0000129762.41889.41; O'Neill JO, 2004, J AM COLL CARDIOL, V44, P820, DOI 10.1016-j.jacc.2004.02.063; Priori SG, 2002, CIRCULATION, V106, P69, DOI 10.1161-01.CIR.0000020013.73106.D8; Priori SG, 2013, EUROPACE, V15, P1389, DOI 10.1093-europace-eut272; Raju H, 2011, HEART RHYTHM, V8, pS41; Roux-Buisson N, 2012, HUM MOL GENET, V21, P2759, DOI 10.1093-hmg-dds104; Steinhaus DA, 2012, AM HEART J, V163, P125, DOI 10.1016-j.ahj.2011.09.016; Sy RW, 2011, CIRCULATION, V124, P2187, DOI 10.1161-CIRCULATIONAHA.111.028258; Tseng ZH, 2009, HEART RHYTHM, V6, P1315, DOI 10.1016-j.hrthm.2009.06.034; van der Werf C, 2010, CIRC-ARRHYTHMIA ELEC, V3, P96, DOI 10.1161-CIRCEP.109.877142; Watanabe J, 2001, CIRCULATION, V104, P1911; Zheng ZJ, 2001, CIRCULATION, V104, P2158, DOI 10.1161-hc4301.098254; Zipes DP, 2006, J AM COLL CARDIOL, V48, pe247, DOI DOI 10.1016-J.JACC.2006.07.0100
Ovarian cancer screening in the general population.
No full textDespite significant improvements in therapy, ovarian cancer continues to be a leading cause of death amongst women with gynaecological malignancies. Advanced stage at diagnosis is thought to be a major contributor to mortality. Hence, there is considerable interest in early detection through screening. In the 1990s, Professor Jacobs pioneered the development of a multimodal ovarian cancer screening (OCS) strategy using serum CA125 as the first line screen and pelvic ultrasound as the second line test. This thesis summarises the next steps in the journey with refining of the screening algorithm, feasibility testing in a pilot randomised control trial (RCT) and finally setting up and recruiting 200,000 women into the largest ever RCT . The risk of ovarian cancer in postmenopausal women with elevated CA125 levels was established through a detailed analysis of 1219 pelvic scans from 741 women with raised CA125 levels in the completed trial of 22,000 women. Based on this, the multimodal 'Risk of Ovarian Cancer' (ROC) algorithm was refined and morphology instead of volume was used to interpret the ovarian scans. The refined ROC algorithm was then prospectively evaluated in a pilot RCT of 13,582 postmenopausal women. The trial established that screening using the ROC algorithm was feasible and could achieve high specificity and positive predictive value. The improved performance characteristics of the screening strategy and the experience accumulated in running and organising the pilot trial led to the design and successful implementation of a RCT - the UK Collaborative Trial of Ovarian Cancer Screening (UKCTOCS) - to assess the impact of early detection on disease mortality. The trial commenced in 2001 with recruitment of 202,638 postmenopausal women by September 2005. The issues involved in setting up the trial, recruitment of 202,000 women and the baseline characteristics of this population are described
