A groundbreaking cognitive science study published in the journal Cognition reveals that the public maintains a firm psychological boundary between artificial intelligence and human consciousness, even when both exhibit identical behaviors. Led by researchers at Ludwig-Maximilians-Universität München (LMU), the study evaluated nearly 1,100 participants to measure how mental states are attributed to AI models versus human actors under identical experimental conditions. While participants readily attributed technical “awareness” to highly responsive AI agents, they consistently refused to credit them with subjective “consciousness”—judging even the most sophisticated AI to be less conscious than the least responsive human. The findings provide critical empirical insights for technology policy, science communication, and regulatory frameworks governing AI deployment.
Disentangling Artificial Behavior from Subjective Consciousness
MUNICH — As artificial intelligence models become increasingly integrated into daily public life—demonstrating sophisticated language generation, goal-oriented decision-making, and seemingly empathetic interactions—a central debate has emerged among ethicists, computer scientists, and policymakers: Do humans genuinely perceive these advanced systems as conscious entities?
A comprehensive empirical study conducted by researchers at Ludwig-Maximilians-Universität München (LMU) and published in the peer-reviewed journal Cognition provides a decisive, nuanced answer. The research demonstrates that while everyday users frequently adopt mentalistic vocabulary to describe AI performance, they maintain a strict, fundamental distinction between human consciousness and artificial processing.
The study, led by Dr. Louis Longin from LMU’s Chair of Philosophy of Mind alongside senior authors Dr. Bahador Bahradori, Professor Ophelia Deroy, and an international collaborative team, represents the first controlled experiment to directly compare human and AI mental state attributions under identical behavioral scenarios.
“Whereas previous studies have typically asked general questions about whether AI actually has mental states, our study is the first to directly compare how people attribute the same mental states to AI and humans behaving in exactly the same way, under identical circumstances,” explained Dr. Longin during a research presentation outlining the project’s methodology.
Methodology and Quantitative Experimental Results
To isolate how behavioral responsiveness and linguistic choices influence public perception, the LMU research team designed a rigorous multi-cohort experimental framework involving nearly 1,100 participants. The experimental architecture divided subjects into controlled reading groups exposed to short, standardized behavioral scenarios.
In the primary experimental arm, one group of participants read descriptions of artificial intelligence agents exhibiting varying degrees of environmental responsiveness—such as reacting to ambient auditory cues or adjusting responses based on the emotional states of nearby individuals. A second control group read precisely identical behavioral scenarios, with the sole modification being that the protagonist was identified as a human being rather than an artificial agent. Following the reading modules, all participants completed standardized evaluation scales measuring the degree of “consciousness” and “awareness” exhibited by the respective protagonist.
The quantitative output revealed a striking divergence between the concepts of functional awareness and internal subjective consciousness:
- The Consciousness Gap: While increased environmental responsiveness led to higher ratings of consciousness for both humans and AI, a persistent, unyielding gap remained between the two categories. Remarkably, even the most responsive, human-like AI agent was consistently rated as less conscious than the least responsive human protagonist.
- The Awareness Convergence: When evaluating “awareness”—a more technical, less philosophically loaded descriptor—the statistical gap between AI and human subjects largely disappeared. Participants attributed environmental awareness to AI agents at rates that directly mirrored human scoring based on observable behavior.
“We may be using mental terms to refer to AI, but this may only be because we lack better words,” noted Professor Ophelia Deroy, philosopher and senior author on the study. Speaking calmly in her academic study in Munich, surrounded by texts on cognitive philosophy, Deroy emphasized the linguistic constraints of modern technology reporting: “After all, AI is trained to act and speak like a human, so these descriptions seem natural.”
Cognitive Mechanics and Linguistic Framing
The research highlights a crucial distinction in human cognitive sociology: the difference between operational vocabulary and ontological belief. While sensational media coverage and marketing materials frequently claim that users are falling into anthropomorphic traps—treating conversational AI as sentient peers—the LMU data shows that public cognition is considerably more sophisticated and discriminating.
Dr. Longin emphasized that public concern over widespread human-AI confusion may be overstated, or at least misdiagnosed.
“There is a growing worry in public discussion that people will see AI behaving in a human-like way and start treating these systems as if they had human-like mental states,” Longin observed. Sitting beside his analytical charts documenting participant responses, he noted: “We found something much more nuanced: People are quite willing to say that an AI notices things and is aware of its surroundings. But they clearly draw a line when it comes to the term ‘consciousness’. People consistently attribute less consciousness to AI than humans – even when their behavior is identical.”
This linguistic divide carries profound implications for how technology corporations, regulatory bodies, and journalists communicate about machine learning advancements. When corporate press releases or news accounts attribute “intentions,” “moral hesitation,” or “plans” to neural networks, they risk distorting functional realities. However, everyday consumers naturally filter these descriptions, recognizing that technical responsiveness does not equate to subjective experience or sentience.
Policy Implications and Industry Science Communication
The findings arrive at a crucial juncture for global technology regulation, as legislative bodies in the United States, European Union, and across Asia draft consumer protection standards surrounding artificial intelligence. Regulators have expressed ongoing concern that anthropomorphic AI interfaces could manipulate vulnerable users by feigning emotional bonds or moral agency.
Dr. Bahador Bahrami, senior co-author and social neuroscience expert, stressed that responsible science communication must align with the cognitive realities revealed by the study.
“Our study teaches us lessons for how we communicate about AI,” Dr. Bahrami stated during an international press briefing discussing the publication. “Companies and journalists often reach for hyped descriptions of AI models, and credit them with intentions, plans, or even moral conscience and hesitation. What our study shows is that everyday judgements are much more discriminating: People do not simply put AI and humans on the same mental scale. But the boundary also depends on which terms we use. For more neutral descriptors such as ‘awareness’, the distinction can largely disappear. That makes our choice of language important.”
As artificial intelligence systems continue to advance in conversational fluency and autonomous task execution, the LMU study provides a reassuring baseline: human observers possess an innate, resilient framework for distinguishing between computational processing and living consciousness. Maintaining precise, accurate terminology in public discourse will remain essential to ensuring that technological progress is understood clearly, without falling prey to unnecessary hype or unfounded alarm.
No Comment! Be the first one.