Submitted:
04 May 2026
Posted:
06 May 2026
You are already at the latest version
Abstract
Artificial intelligence is increasingly embedded in decision-making across organizational and societal contexts, yet it remains unclear whether individuals remain cognitively aligned with decisions generated under algorithmic conditions. Existing research has emphasized trust, fairness, and transparency, but provides limited insight into the cognitive mechanisms that sustain coherent human judgment during system-mediated decision processes.Here we introduce perceptual integrity as a measurable construct capturing the extent to which individuals maintain interpretive coherence and decision authorship in human–AI interaction. We test this framework in a controlled experiment (N = 602) comparing algorithmic imposition with interpretive autonomy. Algorithmic imposition significantly reduced perceptual integrity relative to interpretive autonomy (t(600) = 4.21, p < 0.001, Cohen’s d = 0.38). Perceptual integrity was a significant predictor of trust in AI-assisted decisions (β = 0.36, p < 0.001) and partially mediated the relationship between decision condition and trust (indirect effect = 0.17, 95% CI [0.09, 0.27]).These findings identify perceptual integrity as a cognitive mechanism linking decision structure to trust under system-mediated conditions. More broadly, they suggest that effective integration of algorithmic systems depends not only on performance accuracy but on preserving cognitive alignment during decision formation. This work provides a generalizable framework for understanding how humans remain engaged with decisions in increasingly automated environments.
Keywords:
perceptual integrity
; human–AI interaction
; algorithmic decision-making
; trust in AI
; interpretability
; decision autonomy
; cognitive alignment
; AI-assisted decision-making
Introduction
The rapid expansion of artificial intelligence (AI) across organizational and societal domains has fundamentally transformed how decisions are generated, evaluated, and enacted. From recruitment and performance evaluation to clinical support and strategic planning, algorithmic systems increasingly shape not only decision outcomes but also the processes through which those outcomes are produced. While these systems offer substantial gains in efficiency, scalability, and predictive accuracy, they also introduce a structural shift in which interpretive processes—historically embedded within human cognition—are partially externalized to computational systems [4,5,6].
A growing body of research has examined human responses to AI-assisted decision-making, primarily through constructs such as trust, fairness, and transparency [1,4,10]. Studies on algorithm aversion and appreciation demonstrate that individuals may either reject or favor algorithmic recommendations depending on contextual cues and prior experience [2,3,9]. More recent work highlights the risk of overreliance on automated systems, where individuals defer to algorithmic outputs at the expense of independent reasoning [7,8]. Collectively, these findings indicate that human–AI interaction is not merely a technical problem but a cognitive and behavioral one [4,7,8].
Despite these advances, existing frameworks remain largely outcome-oriented. Constructs such as trust and perceived fairness explain whether individuals accept or reject AI-assisted decisions, but they provide limited insight into how individuals experience the decision-making process itself. In particular, current literature does not adequately address whether individuals remain cognitively aligned with decisions that are partially or fully generated by algorithmic systems. This distinction is critical, as decision acceptance does not necessarily imply decision authorship; individuals may comply with or even trust algorithmic outputs while experiencing reduced interpretive coherence with the underlying decision process [9,18].
This gap points to a deeper, underexplored dimension of human–AI interaction: the integrity of the perceptual process through which decisions are formed. In environments where decision authority is distributed between human and artificial agents, maintaining alignment between interpretation, meaning, and action becomes a central challenge. Without such alignment, decision-making may appear efficient at the system level while becoming fragmented at the cognitive level, with potential consequences for trust, engagement, and long-term effectiveness [5,10,13]. Emerging evidence further suggests that the interpretability of AI systems plays a critical role in shaping user understanding, trust calibration, and decision engagement [18,19].
To address this limitation, the present study introduces perceptual integrity as a construct capturing the extent to which individuals maintain coherence between their interpretation of a situation, the decision process, and the resulting outcome. Unlike trust, which reflects an evaluation of the system, or autonomy, which reflects perceived control, perceptual integrity focuses on the preservation of cognitive authorship during decision formation. As such, it represents a more foundational condition shaping how individuals engage with and internalize algorithmically mediated decisions.
This study builds on a systems-based perspective of cognition in which decision-making emerges from the interaction among environmental context, memory, system input, and human agency. Within this framework, decision quality is not determined solely by accuracy or efficiency but by the degree of balance among these interacting elements. Perceptual integrity is positioned as the measurable manifestation of this balance, reflecting whether individuals remain cognitively engaged and interpretively aligned within system-driven environments.
Empirically, we examine how different decision structures—specifically, algorithmic imposition versus interpretive autonomy—affect perceptual integrity and how perceptual integrity, in turn, influences trust. Using a randomized controlled experimental design, the study tests whether the structure of decision authority shapes cognitive alignment and whether this alignment functions as a mechanism linking decision processes to trust outcomes.
This study makes three primary contributions. First, it advances research on human–AI interaction by introducing perceptual integrity as a construct that captures decision coherence rather than decision acceptance. Second, it extends existing perspectives on decision-making and leadership by shifting the focus from behavioral outcomes to the management of cognitive conditions in technologically mediated environments. Third, it provides empirical evidence that interpretive alignment represents a key mechanism through which algorithmic decision structures influence psychological outcomes.
By reframing the problem from optimization to cognitive coherence, this work contributes to a more comprehensive understanding of human–AI systems. As algorithmic infrastructures continue to expand, the central question is no longer only whether systems produce better decisions, but whether individuals remain cognitively present within the decisions those systems help produce.
Study Design
We employed a randomized controlled experimental design to examine the causal effects of decision structure on perceptual integrity and trust in a simulated human–AI decision-making context. Participants were randomly assigned to one of two conditions: (1) algorithmic imposition, in which participants were required to follow an AI-generated recommendation, and (2) interpretive autonomy, in which participants were allowed to accept, modify, or override the recommendation. This design isolates the effect of decision authority while holding informational input constant, enabling causal inference regarding the cognitive consequences of algorithmic control.
Participants
A total of 602 participants were recruited via Prolific to ensure demographic diversity and data quality. Eligibility criteria included age ≥18 years and fluency in English. Participants represented diverse professional backgrounds, including business, healthcare, and technology.
The sample ranged in age from 21 to 55 years (M = 34.2, SD = 8.7) with balanced gender representation. All participants reported prior exposure to digital systems, and the majority indicated familiarity with AI-assisted environments.
The study was conducted in accordance with institutional ethical standards and approved by the relevant Institutional Review Board. Informed consent was obtained from all participants, and data were collected anonymously.
Procedure
The experiment was administered online using a structured survey interface. Participants completed a standardized decision-making task involving the selection of a project manager from four candidates. Each candidate profile included multiple dimensions, including performance metrics, peer evaluations, and leadership competencies.
An AI-generated recommendation was provided in both conditions based on a predefined scoring algorithm integrating quantitative and qualitative indicators. The recommendation was accompanied by a brief justification to reflect typical decision-support systems.
In the algorithmic imposition condition, participants were required to follow the recommendation, and alternative selections were restricted. In the interpretive autonomy condition, participants were allowed to accept, modify, or override the recommendation, preserving interpretive engagement.
Manipulation check
To verify the effectiveness of the manipulation, participants completed a three-item perceived autonomy scale immediately following the task (e.g., “I had the ability to influence the final decision”). Scores differed significantly between conditions (p < 0.001), confirming that the interpretive autonomy condition induced greater perceived control.
Measures
Perceptual integrity
Perceptual integrity was operationalized as a multi-item construct capturing coherence between interpretation, decision process, and outcome. The scale comprised 21 items across three dimensions: interpretive coherence, system alignment, and temporal continuity. Responses were recorded on a 5-point Likert scale (1 = strongly disagree, 5 = strongly agree). The scale demonstrated high internal consistency (Cronbach’s α = 0.91).
Trust
Trust was measured using a 4-item scale assessing confidence in the decision outcome and perceived reliability of the process. Responses were recorded on a 5-point Likert scale. Internal consistency was satisfactory (Cronbach’s α = 0.87).
Control variables
Analyses controlled for demographic variables (age, gender, education) and prior familiarity with AI systems, given their documented influence on perceptions of algorithmic decision-making.
Statistical analysis
Independent samples t-tests were used to assess differences between experimental conditions. Multiple regression analysis was conducted to evaluate the relationship between perceptual integrity and trust. Mediation analysis was performed using PROCESS Model 4 with 5,000 bootstrap samples to estimate indirect effects and corresponding confidence intervals. Statistical significance was evaluated at α = 0.05. Analyses were conducted using standard statistical software (SPSS or R).
Data availability
Data supporting the findings of this study are available from the corresponding author upon reasonable request.
Results
Following data screening, responses that failed the manipulation check were excluded (n = 27), resulting in a final sample of 575 participants. The manipulation was effective: participants in the interpretive autonomy condition reported higher perceived control (M = 4.12, SD = 0.76) than those in the algorithmic imposition condition (M = 2.98, SD = 0.89; t(573) = 10.21, p < 0.001). Overall, perceptual integrity (M = 3.47, SD = 0.72) and trust (M = 3.58, SD = 0.69) showed moderate variability, suggesting heterogeneity in participants’ responses to AI-assisted decision-making.
Decision structure was associated with differences in perceptual integrity. Participants in the algorithmic imposition condition reported lower perceptual integrity (M = 3.28, SD = 0.71) than those in the interpretive autonomy condition (M = 3.65, SD = 0.68; t(573) = 4.21, p < 0.001, Cohen’s d = 0.38). Although statistically reliable, the effect size was moderate, indicating that decision structure influences—but does not fully determine—perceived coherence in decision-making.
Perceptual integrity was positively associated with trust. Regression analysis indicated that higher perceptual integrity predicted higher trust in AI-assisted decisions (β = 0.36, SE = 0.06, t = 6.02, p < 0.001), accounting for a modest proportion of variance (R² = 0.16). This suggests that perceptual integrity represents one of several factors contributing to trust formation.
Mediation analysis indicated a significant indirect effect of decision condition on trust through perceptual integrity (indirect effect = 0.17, SE = 0.04, 95% CI [0.09, 0.27]). The direct effect remained significant (β = 0.18, SE = 0.07, p = 0.02), consistent with partial mediation. These findings suggest that perceptual integrity accounts for part of the relationship between decision structure and trust, while additional mechanisms are likely involved.
Correlation analyses provided evidence for discriminant validity. Perceptual integrity showed moderate associations with perceived autonomy (r = 0.49) and fairness (r = 0.44), and a somewhat stronger association with trust (r = 0.51). All correlations were below 0.70, supporting the distinctiveness of the construct.
Including demographic variables (age, gender, education) and prior familiarity with AI did not materially change the results. The pattern of findings remained consistent, suggesting that the observed effects are not solely attributable to demographic variation.
Discussion
The present study moves beyond outcome-based accounts of human–AI interaction by examining the cognitive conditions under which individuals remain coherent participants in system-assisted decision-making. While prior research has emphasized trust, fairness, and transparency [1,4,10], the present findings suggest that these constructs depend, in part, on a more fundamental process—whether individuals retain interpretive alignment and a sense of authorship within the decision itself.
The results provide consistent evidence that decision structure influences perceptual integrity. Participants exposed to algorithmic imposition reported lower levels of perceptual integrity compared to those allowed interpretive autonomy, indicating that restricting engagement with the decision process reduces perceived coherence. Importantly, this effect was moderate, suggesting that algorithmic systems influence—but do not fully determine—human cognition. This aligns with emerging perspectives that human–AI interaction involves negotiation rather than replacement of human judgment [5,13,14].
A central contribution of this study lies in demonstrating that perceptual integrity is a meaningful predictor of trust. Unlike traditional approaches that conceptualize trust as a response to system performance, the present findings indicate that trust is also shaped by the individual’s cognitive experience of the decision process. Individuals were more likely to trust outcomes when they perceived their decisions as coherent and interpretable. This helps explain inconsistencies in prior research on algorithm aversion and appreciation [2,3,9], where acceptance of AI does not consistently align with objective system accuracy.
The mediation analysis further supports this interpretation by showing that perceptual integrity partially explains the relationship between decision structure and trust. This suggests that algorithmic systems influence trust not only directly, but through their impact on how individuals experience decision formation. However, the partial mediation observed indicates that perceptual integrity operates alongside other mechanisms, such as perceived competence, fairness, or outcome expectations [4,10], highlighting the multifactorial nature of trust in human–AI contexts.
Taken together, these findings suggest that existing constructs are insufficient to fully explain human responses to AI-assisted decision-making. While trust captures whether individuals accept or rely on a system, it does not capture whether individuals remain cognitively aligned with the decisions produced. In this sense, perceptual integrity reflects a distinct dimension—decision authorship—capturing whether individuals experience the decision as their own. This distinction is critical, as individuals may comply with or trust algorithmic outputs while experiencing reduced cognitive ownership or understanding [9,18].
From a theoretical perspective, these results support a shift from outcome-based to process-based models of human–AI interaction. The findings suggest that decision quality cannot be understood solely in terms of accuracy or efficiency, but must also consider the integrity of the cognitive process through which decisions are formed. This perspective is consistent with recent work emphasizing the importance of interpretability and cognitive engagement in human–AI systems [18,19].
This perspective also extends existing views on leadership in technologically mediated environments. Rather than focusing solely on behavioral outcomes or performance optimization, leadership can be reconceptualized as the management of cognitive conditions that enable coherent decision-making. In contexts where algorithmic systems increasingly structure decisions, maintaining interpretive space becomes a critical function. This aligns with emerging research on collective cognition and AI-mediated collaboration [14,15].
The findings contribute to what we term a framework of conscious leadership, in which leadership is understood as the preservation of cognitive balance in environments shaped by intelligent systems. Within this framework, perceptual integrity represents a central mechanism capturing whether individuals remain active participants in their own decision processes.
Limitations and Future Research
Several limitations should be acknowledged. First, the study employed a controlled experimental design, which, while enabling causal inference, may not fully capture the complexity of real-world decision environments. Future research should examine perceptual integrity in organizational and longitudinal contexts to assess its stability and ecological validity.
Second, perceptual integrity was measured using self-report instruments, which may be subject to subjective bias. Future work could incorporate behavioral and process-based measures—such as response time, decision revision patterns, or eye-tracking—to more directly capture cognitive engagement.
Third, the study focused on a general decision-making scenario. The dynamics of perceptual integrity may differ across domains, particularly in high-stakes environments such as healthcare, finance, or public policy, where decision consequences are more consequential.
Finally, although the proposed framework provides a conceptual foundation, further empirical research is required to validate and refine the relationships among perceptual integrity and related constructs. In particular, future studies could examine moderating factors such as expertise, system transparency, and task complexity, which have been shown to influence human–AI interaction outcomes [7,8].
Conclusions
This study identifies perceptual integrity as a cognitive mechanism linking decision structure to trust under algorithmic conditions. Rather than treating trust as a direct response to system performance alone, the findings indicate that trust depends, in part, on whether individuals remain cognitively aligned with the decisions they enact.
Across a controlled experimental design, restricting interpretive engagement reduced perceptual integrity, and this reduction was associated with lower trust. The observed effects were moderate and partially mediated, suggesting that perceptual integrity represents one component within a broader set of mechanisms shaping human responses to AI-assisted decision-making.
These findings shift attention from outcome-based evaluations of algorithmic systems toward the cognitive conditions under which decisions are formed and experienced. As algorithmic infrastructures continue to expand, preserving cognitive alignment—through interpretability, engagement, and meaningful decision participation—may be as critical as improving predictive accuracy.
By reframing human–AI interaction around cognitive coherence rather than performance alone, this work provides a foundation for designing decision systems that support not only better outcomes, but more coherent human participation within those outcomes.
Author Contributions
Abdulmohsen H. Alrohaimi conceived the study, developed the methodology, conducted the analysis, and wrote the manuscript. Abdulmohsen H. Alrohaimi reviewed and approved the final version of the manuscript.
Funding
This research received no external funding.
Institutional Review Board Statement
The study was conducted in accordance with relevant institutional guidelines. Informed consent was obtained from all participants.
Acknowledgments
The author acknowledges institutional support from Shaqra University. Language editing support was assisted using a large language model (ChatGPT, OpenAI). All outputs were reviewed and validated by the author.
Data availability
Data supporting the findings of this study are available from the corresponding author upon reasonable request.
Conflicts of Interest
The author declares no conflicts of interest.
References
- Mayer, R.C.; Davis, J.H.; Schoorman, F.D. An integrative model of organizational trust. Acad. Manag. Rev. 1995, 20, 709–734. [Google Scholar] [CrossRef]
- Dietvorst, B.J.; Simmons, J.P.; Massey, C. Algorithm aversion: People erroneously avoid algorithms after seeing them err. J. Exp. Psychol. General. 2015, 144, 114–126. [Google Scholar] [CrossRef] [PubMed]
- Logg, J.M.; Minson, J.A.; Moore, D.A. Algorithm appreciation: People prefer algorithmic to human judgment. Organ. Behav. Hum. Decis. Process. 2019, 151, 90–103. [Google Scholar] [CrossRef]
- Burton, J.W.; Stein, M.-K.; Jensen, T.B. A systematic review of algorithm aversion. J. Behav. Decis. Mak. 2020, 33, 220–239. [Google Scholar] [CrossRef]
- Glikson, E.; Woolley, A.W. Human trust in artificial intelligence: Review of empirical research. Acad. Manag. Ann. 2020, 14, 627–660. [Google Scholar] [CrossRef]
- Raisch, S.; Krakowski, S. Artificial intelligence and management: The automation–augmentation paradox. Acad. Manag. Rev. 2021, 46, 192–210. [Google Scholar] [CrossRef]
- Jussupow, E.; Spohrer, K.; Heinzl, A.; Barrot, C. Augmenting medical diagnosis decisions? Inf. Syst. Res. 2021, 32, 713–735. [Google Scholar] [CrossRef]
- Buçinca, Z.; Malaya, M.B.; Gajos, K.Z. To trust or to think: Cognitive forcing functions reduce overreliance on AI. CHI Conference on Human Factors in Computing Systems, 2021. [Google Scholar] [CrossRef]
- Bansal, G.; Wu, T.; Zhou, D.; Fok, R. Does the whole exceed its parts? Effects of AI explanations on complementary team performance. In CHI Conference on Human Factors in Computing Systems; 2021. [Google Scholar] [CrossRef]
- Köbis, N.; Bonnefon, J.-F.; Rahwan, I. Bad machines corrupt good morals. Nat. Hum. Behav. 2021, 5, 679–685. [Google Scholar] [CrossRef] [PubMed]
- Siau, K.; Wang, W. Artificial intelligence ethics: Ethics of AI and ethical AI. J. Database Manag. 2020, 31, 74–87. [Google Scholar] [CrossRef]
- Dellermann, D.; Ebel, P.; Söllner, M.; Leimeister, J.M. Hybrid intelligence. Bus. Inf. Syst. Eng. 2019, 61, 637–643. [Google Scholar] [CrossRef]
- Faraj, S.; Pachidi, S.; Sayegh, K. Working and organizing in the age of artificial intelligence. Inf. Organ. 2018, 28, 62–70. [Google Scholar] [CrossRef]
- Schmutz, J.B.; Outland, N.; Kerstan, S.; Georganta, E.; Ulfert, A.-S. AI-teaming: Redefining collaboration in the digital era. Curr. Opin. Psychol. 2024, 58, 101837. [Google Scholar] [CrossRef] [PubMed]
- Woolley, A.W.; Gupta, P. Understanding collective intelligence. Perspect. Psychol. Sci. 2024, 19, 344–354. [Google Scholar] [CrossRef] [PubMed]
- van Knippenberg, D.; Pearce, C.L.; van Ginkel, W.P. Shared leadership–vertical leadership dynamics in teams. Group Organ. Manag. 2025, 50, 44–67. [Google Scholar] [CrossRef]
- De Vincenzo, F.; Curșeu, P.L.; Chirilă, M. Collective leadership and team cognition: A systematic review. Acta Psychol. 2025, 259, 105403. [Google Scholar] [CrossRef]
- Ribeiro, M.T.; Singh, S.; Guestrin, C. “Why should I trust you?” Explaining the predictions of any classifier. In KDD; 2016. [Google Scholar] [CrossRef]
- Lai, V.; Tan, C. On human predictions with explanations and predictions of machine learning models. In FAT*; 2019. [Google Scholar] [CrossRef]
- Bansal, G.; Nushi, B.; Kamar, E.; et al. Beyond accuracy: The role of mental models in human-AI teaming. In AAAI HCOMP; 2019. [Google Scholar]
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content. |
© 2026 by the author. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license.
Copyright: This open access article is published under a Creative Commons CC BY 4.0 license, which permit the free download, distribution, and reuse, provided that the author and preprint are cited in any reuse.
