Determinants of Judgments of Explanatory Power: Credibility, Generality, and Statistical Relevance

Colombo, Matteo; Bucher, Leandra; Sprenger, Jan

doi:10.3389/fpsyg.2017.01430

Cited by 16 publications

(13 citation statements)

References 56 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Would infants' interest in these explanations, and their change in 582 subsequent exploratory behavior after seeing these explanations, decrease as a function of the 583 explanations' goodness? Adults are sensitive to such variations in "explanatory power," where 584 information is said to have more explanatory power as it renders observations less and less 585 surprising (Colombo, Bucher, & Sprenger, 2017;Schupbach & Sprenger, 2011). Older children 586 too are sensitive to explanatory power, in that they recognize explanations that, because of their 587 circularity, are wholly unsatisfying (Baum, Danovitch, & Keil, 2008;Castelain, Bernard, & 588 Mercier, 2018).…”

Section: Results 463mentioning

confidence: 99%

Violations of expectation trigger infants to search for explanations

Perez¹,

Feigenson²

2020

Preprint

View full text Add to dashboard Cite

Infants look longer and explore more following violations-of-expectation, but the reasons for these surprise-induced behaviors are unclear. One possibility is that expectancy violations increase arousal generally, thereby increasing infants’ post-surprise activity. Another possibility is that infants’ exploration reflects the search for an explanation for the surprising event. We tested these alternatives in three experiments. First in Experiment 1 we confirmed that seeing an object violate expectations (by passing through a solid wall) increased infants’ exploration of the surprising object compared to a novel object, and compared to when no expectancy violation was seen. Then in Experiment 2 we measured infants’ exploration after they saw the same violation event, but then an explanation for the event was provided (the wall was revealed to have a large hole in it). We found that providing this explanation abolished infants’ surprise-induced exploration. Furthermore, the longer infants looked at the explanation, the greater their reversal in exploratory preference (i.e., the more they ignored the surprising object and explored the novel object). In Experiment 3 we replicated these effects. These findings demonstrate that preverbal infants both seek and recognize explanations for surprising events.

show abstract

Section: Results 463mentioning

confidence: 99%

Violations of expectation trigger infants to search for explanations

Perez¹,

Feigenson²

2020

Preprint

View full text Add to dashboard Cite

show abstract

“…However, some found that too much explanation can create confusion and degrade trust, and thus argue that AI explanations should only surface the most relevant information [13,16,29]. According to psychological research, humans prefer explanations that are simpler, more probable, and more causally relevant, all else being equal [4,24].…”

Section: Related Workmentioning

confidence: 99%

The effects of example-based explanations in a machine learning interface

Cai

Jongejan

Holbrook

2019

Proceedings of the 24th International Conference on Intelligent User Interfaces

150

View full text Add to dashboard Cite

The black-box nature of machine learning algorithms can make their predictions difficult to understand and explain to end-users. In this paper, we propose and evaluate two kinds of example-based explanations in the visual domain, normative explanations and comparative explanations (Figure 1), which automatically surface examples from the training set of a deep neural net sketch-recognition algorithm. To investigate their effects, we deployed these explanations to 1150 users on QuickDraw, an online platform where users draw images and see whether a recognizer has correctly guessed the intended drawing. When the algorithm failed to recognize the drawing, those who received normative explanations felt they had a better understanding of the system, and perceived the system to have higher capability. However, comparative explanations did not always improve perceptions of the algorithm, possibly because they sometimes exposed limitations of the algorithm and may have led to surprise. These findings suggest that examples can serve as a vehicle for explaining algorithmic behavior, but point to relative advantages and disadvantages of using different kinds of examples, depending on the goal.

show abstract

“…The findings of the current prospective study are limited as data has been collected from only one cardiac rehabilitation centre. Accordingly, the generalisability of these findings might be limited (Colombo, Bucher, & Sprenger, 2017). In addition, it is well established that small studies may lack power, making it difficult to refute the proposed relationships (Wasserstein & Lazar, 2016; Greenland et al, 2016).…”

Section: Discussionmentioning

confidence: 99%

The Process of Patient Engagement in Outpatient Cardiac Rehabilitation Programs

et al. 2019

View full text Add to dashboard Cite

The primary aim of this study was to test the causal structure of the model of therapeutic engagement (MTE) for the first time, to examine whether the model assists in understanding the process of patient engagement in cardiac rehabilitation (CR) programs. This study used a prospective design, following up patients from the Gold Coast University Hospital Cardiology ward who attended Robina Cardiac Rehabilitation Clinic. A structural equation model of the interactions among the proposed variables within the three stages of the MTE (intention to engage in CR programs, CR initiation, and sustained engagement) revealed significant relationships among these variables in a dataset of 101 patients who attended a CR program. However, no relationship was discerned between outcome expectancies and patient intention to engage in CR. Patients’ willingness to consider the treatment also mediated the relationship between perceived self-efficacy and patient intention to engage in CR. These findings help clarify the process proposed by Lequerica and Kortte (2010) in the context of patient engagement in CR programs. The findings also reveal information on how patients engage in CR programs. Importantly, this provides new information for healthcare providers, enabling them to more effectively engage patients according to their stage of engagement.

show abstract

Determinants of Judgments of Explanatory Power: Credibility, Generality, and Statistical Relevance

Cited by 16 publications

References 56 publications

Violations of expectation trigger infants to search for explanations

Violations of expectation trigger infants to search for explanations

The effects of example-based explanations in a machine learning interface

The Process of Patient Engagement in Outpatient Cardiac Rehabilitation Programs

Contact Info

Product

Resources

About