Felix Kares
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Felix Kares has a total of 11 co-authors. Here are the top co-authors by number of shared publications:
- Markus Langer
University of Freiburg
3 shared publications - Cornelius J. König
Saarland University
2 shared publications - Denise Vesper
Department of Psychology, Universität des Saarlandes, Saarbrücken, Germany
1 shared publication - Hanwei Zhang
Saarland University, Saarland Informatics Campus
1 shared publication - Hanwei Zhang
Department of Computer Science, Saarland University, Saarbrücken, Germany
1 shared publication Leon Hannig
University of Duisburg-Essen
1 shared publication
Latest Publications
- Open Access
- 2 September 2026
- Published by Elsevier in Computers in Human Behavior
- Vol. 187
Are explainable AI (XAI) evaluation strategies aligned? Comparing subjective, objective, and mathematical evaluation measures using saliency maps
- 2 September 2026
- Published by Elsevier in Computers in Human Behavior
- Vol. 187
Open AccessAre explainable AI (XAI) evaluation strategies aligned? Comparing subjective, objective, and mathematical evaluation measures using saliency maps
- Open Access
- 4 August 2026
- Published by Hogrefe Publishing Group in European Journal of Psychology Open
Seeing What You Believe
Introduction: Political ideology can shape how individuals evaluate information, including scientific research. Aim: This preregistered study examined whether psychology undergraduates ( N = 723) evaluated research credibility based on political alignment and methodological quality. Methods: Using a 2 (Political leaning of findings: left-leaning vs. right-leaning) × 2 (Methodological quality: high vs. low) between-subjects design, participants rated the credibility of fictional study summaries. Results: Results showed that psychology students were significantly more left-leaning than a representative sample of young adults and rated politically congruent research as more credible, indicating confirmation bias. However, methodological rigor remained a strong and consistent predictor of perceived credibility, and no three-way interaction emerged between participant ideology, political alignment, and methodological quality. Conclusion: These findings suggest that, although political biases in scientific evaluation exist, critical engagement with research methods remains a strong predictor of credibility. Future research should examine whether this holds true for other groups.
- 4 August 2026
- Published by Hogrefe Publishing Group in European Journal of Psychology Open
Open AccessSeeing What You Believe
Introduction: Political ideology can shape how individuals evaluate information, including scientific research. Aim: This preregistered study examined whether psychology undergraduates ( N = 723) evaluated research credibility based on political alignment and methodological quality. Methods: Using a 2 (Political leaning of findings: left-leaning vs. right-leaning) × 2 (Methodological quality: high vs. low) between-subjects design, participants rated the credibility of fictional study summaries. Results: Results showed that psychology students were significantly more left-leaning than a representative sample of young adults and rated politically congruent research as more credible, indicating confirmation bias. However, methodological rigor remained a strong and consistent predictor of perceived credibility, and no three-way interaction emerged between participant ideology, political alignment, and methodological quality. Conclusion: These findings suggest that, although political biases in scientific evaluation exist, critical engagement with research methods remains a strong predictor of credibility. Future research should examine whether this holds true for other groups.
- Open Access
- 24 June 2026
- Published by Elsevier in Computers in Human Behavior: Artificial Humans
- Vol. 9
Responsibility attributions in AI-assisted decisions: The interplay of the level of automation and system reliability
- 24 June 2026
- Published by Elsevier in Computers in Human Behavior: Artificial Humans
- Vol. 9
Open AccessResponsibility attributions in AI-assisted decisions: The interplay of the level of automation and system reliability
- Open Access
- 5 May 2026
- Published by Taylor & Francis in Behaviour & Information Technology
- p. 1-17
Working with Large Language Models: impacts on performance, efficiency and perceived responsibility
1. The use of Large Language Models (LLMs) transforms modern workplaces with key goals being to enhance human performance and efficiency (e.g. Eloundou et al. 2024). LLMs can also accelerate learni...
- 5 May 2026
- Published by Taylor & Francis in Behaviour & Information Technology
- p. 1-17
Open AccessWorking with Large Language Models: impacts on performance, efficiency and perceived responsibility
1. The use of Large Language Models (LLMs) transforms modern workplaces with key goals being to enhance human performance and efficiency (e.g. Eloundou et al. 2024). LLMs can also accelerate learni...
- Open Access
What Makes for a Good Saliency Map? Comparing Strategies for Evaluating Saliency Maps in Explainable AI (XAI)
Saliency maps are a popular approach for explaining classifications of (convolutional) neural networks. However, it remains an open question as to how best to evaluate salience maps, with three families of evaluation methods commonly being used: subjective user measures, objective user measures, and mathematical metrics. We examine three of the most popular saliency map approaches (viz., LIME, Grad-CAM, and Guided Backpropagation) in a between subject study (N=166) across these families of evaluation methods. We test 1) for subjective measures, if the maps differ with respect to user trust and satisfaction; 2) for objective measures, if the maps increase users' abilities and thus understanding of a model; 3) for mathematical metrics, which map achieves the best ratings across metrics; and 4) whether the mathematical metrics can be associated with objective user measures. To our knowledge, our study is the first to compare several salience maps across all these evaluation methods$-$with the finding that they do not agree in their assessment (i.e., there was no difference concerning trust and satisfaction, Grad-CAM improved users' abilities best, and Guided Backpropagation had the most favorable mathematical metrics). Additionally, we show that some mathematical metrics were associated with user understanding, although this relationship was often counterintuitive. We discuss these findings in light of general debates concerning the complementary use of user studies and mathematical metrics in the evaluation of explainable AI (XAI) approaches.
Open AccessWhat Makes for a Good Saliency Map? Comparing Strategies for Evaluating Saliency Maps in Explainable AI (XAI)
Saliency maps are a popular approach for explaining classifications of (convolutional) neural networks. However, it remains an open question as to how best to evaluate salience maps, with three families of evaluation methods commonly being used: subjective user measures, objective user measures, and mathematical metrics. We examine three of the most popular saliency map approaches (viz., LIME, Grad-CAM, and Guided Backpropagation) in a between subject study (N=166) across these families of evaluation methods. We test 1) for subjective measures, if the maps differ with respect to user trust and satisfaction; 2) for objective measures, if the maps increase users' abilities and thus understanding of a model; 3) for mathematical metrics, which map achieves the best ratings across metrics; and 4) whether the mathematical metrics can be associated with objective user measures. To our knowledge, our study is the first to compare several salience maps across all these evaluation methods$-$with the finding that they do not agree in their assessment (i.e., there was no difference concerning trust and satisfaction, Grad-CAM improved users' abilities best, and Guided Backpropagation had the most favorable mathematical metrics). Additionally, we show that some mathematical metrics were associated with user understanding, although this relationship was often counterintuitive. We discuss these findings in light of general debates concerning the complementary use of user studies and mathematical metrics in the evaluation of explainable AI (XAI) approaches.