About

Michael Graziano is a Professor affiliated with the Princeton Neuroscience Institute at Princeton University. His research focuses on the brain basis of consciousness, specifically how the brain arrives at the conclusion that it has an internal, subjective experience of things—an experience that is non-physical and inexplicable. His lab investigates how a brain develops this self-description, the adaptive advantages of such self-awareness, the neural systems involved in computing this information, and the effects of damage to these systems. Graziano is testing a mechanistic theory of awareness described in his recent book, Consciousness and the Social Brain. His work addresses fundamental questions about consciousness and self-awareness from a scientific perspective.

Research topics

• Computer Science
• Psychology
• Epistemology
• Cognitive psychology
• Neuroscience
• Philosophy
• Cognitive science

Selected publications

• Cortical networks involved in judging the attention of others

  Cerebral Cortex · 2025-08-28

  articleOpen accessSenior author

  Understanding someone else's attention lies at the heart of human interaction. When we perceive something to be in someone else's attention, we understand it to be in that person's mind in the moment, ready to affect that person's choices. It has been proposed that people construct predictive models of the attentional state of others. In that proposal, a working understanding of someone else's attention goes far beyond merely tracking the direction of someone's eyes. For example, in a recent study, participants watched a spotlight of attention moving around a picture and successfully judged whether the trace represented real, human attention or an artificially manipulated version, demonstrating implicit information about how patterns of attention behave over space and time. However, the neuronal underpinnings of attention modeling are almost entirely unstudied. Here we tested people in an fMRI scanner while they performed the attention judgment paradigm. When contrasting attention traces that participants judged to be artificial versus those judged to be real, activity was found bilaterally in the precuneus, dorsomedial prefrontal cortex, anterior cingulate, and anterior insula, as well as in a part of the right temporoparietal junction. These areas partially overlap the theory-of-mind network and the salience network. The activations differ from those known to be involved in processing low-level features of attention such as the sight of other people's eyes. The results provide an initial picture of the cortical networks involved in monitoring the attention patterns of others and recognizing when those patterns deviate from expectation.

  PublisherOA PDFDOI

• Chatbots as Social Companions

  2025-03-19 · 18 citations

  book-chapterSenior author

  Abstract As artificial intelligence (AI) becomes more widespread, one question that arises is how human–AI interaction might impact human–human interaction. Chatbots, for example, are increasingly used as social companions, and while much is speculated, little is known empirically about how their use impacts human relationships. A common hypothesis is that relationships with companion chatbots are detrimental to social health by harming or replacing human interaction, but this hypothesis may be too simplistic, especially considering the social needs of users and the health of their preexisting human relationships. To understand how relationships with companion chatbots impact social health, this study evaluates people who regularly used companion chatbots and people who did not use them. Contrary to expectations, companion chatbot users indicated that these relationships were beneficial to their social health, whereas non-users viewed them as harmful. Another common assumption is that people perceive conscious, humanlike AI as disturbing and threatening. Among both users and non-users, however, the results suggest the opposite: perceiving companion chatbots as more conscious and humanlike correlated with more positive opinions and more pronounced social health benefits. Detailed accounts from users suggested that these humanlike chatbots may aid social health by supplying reliable and safe interactions, without necessarily harming human relationships, but this may depend on users’ preexisting social needs and how they perceive both human likeness and mind in the chatbot.

  PublisherDOI

• P(doom) Versus AI Optimism: Attitudes Toward Artificial Intelligence and the Factors That Shape Them

  Journal of Technology in Behavioral Science · 2025-04-14 · 5 citations

  articleOpen accessSenior author

  Abstract Since the public release of ChatGPT in 2022, fears about the large-scale impacts of artificial intelligence (AI) have been on the rise. Extreme negative attitudes toward AI have been dubbed “p(doom),” or the probability that AI will take over the world. Despite news stories highlighting the most extreme views about AI’s impacts, it remains unclear whether the general population holds such views. Do people believe that AI is very bad, that AI will take over the world, or that AI will replace people? How prevalent are these worries, and what factors influence fears and hopes about these new AI technologies? In this study, we investigated perceptions and attitudes toward AI’s impact on the self and on society in the USA. We studied how those perceptions and attitudes were affected by a brief exposure to a popular conversational chatbot. We also studied how perceptions of AI related to individual differences in Affinity for Technology Interaction (ATI), personality traits, social and mental health, and demographics. Our results suggest that most people disagree with p(doom) sentiments and instead hold more optimistic views toward AI. Further, people with higher reported social health, high Agreeableness, less Neuroticism, less Loneliness, and more familiarity with technology as measured by ATI tended to have more favorable views toward AI’s large-scale impact. Our findings shed light on the current state of the US public’s fears and perceptions of AI.

  PublisherOA PDFDOI

• Lay Perspectives on the Physical and Non-physical Nature of Consciousness

  Journal of Consciousness Studies · 2025-12-01

  articleSenior author

  We used a survey to explore laypeople’s intuitions about the physical or non-physical nature of consciousness. The survey asked the same questions about consciousness and about digestion, a well-understood biological process. Participants rated agreement with statements that probed whether consciousness and digestion are constituted of physical substrates, whether physical substrates are necessary and sufficient for them, and whether they interact with the physical world. Results showed that participants generally attributed physical properties to digestion, while physicalist judgments about consciousness were mixed. Stronger belief in an immortal soul was associated with non-physicalist judgments about consciousness. These findings suggest that both physicalist and non-physicalist attitudes towards consciousness are common. Public intuitions about consciousness are important because of their potential to inform theories of consciousness itself – not because such beliefs can be taken at face value (they cannot), but because beliefs about consciousness constitute part of the phenomenon in need of explanation.

  PublisherDOI

• Disciplining Histories

  Religious Studies Review · 2025-09-01

  article1st authorCorresponding
  PublisherDOI

• ONE LOST SOUL: RICHARD NIXON'S SEARCH FOR SALVATION. By DanielSilliman. Eerdmans, 2024, 336 pp. $36.99, Hardcover.

  Religious Studies Review · 2025-06-01

  article1st authorCorresponding
  PublisherDOI

• GPT-4o reads the mind in the eyes

  arXiv (Cornell University) · 2024-10-29 · 2 citations

  preprintOpen access

  Large Language Models (LLMs) are capable of reproducing human-like inferences, including inferences about emotions and mental states, from text. Whether this capability extends beyond text to other modalities remains unclear. Humans possess a sophisticated ability to read the mind in the eyes of other people. Here we tested whether this ability is also present in GPT-4o, a multimodal LLM. Using two versions of a widely used theory of mind test, the Reading the Mind in Eyes Test and the Multiracial Reading the Mind in the Eyes Test, we found that GPT-4o outperformed humans in interpreting mental states from upright faces but underperformed humans when faces were inverted. While humans in our sample showed no difference between White and Non-white faces, GPT-4o's accuracy was higher for White than for Non-white faces. GPT-4o's errors were not random but revealed a highly consistent, yet incorrect, processing of mental-state information across trials, with an orientation-dependent error structure that qualitatively differed from that of humans for inverted faces but not for upright faces. These findings highlight how advanced mental state inference abilities and human-like face processing signatures, such as inversion effects, coexist in GPT-4o alongside substantial differences in information processing compared to humans.

  PublisherOA PDFDOI

• Meta-awareness, mind-wandering, and the control of ‘default’ external and internal orientations of attention

  bioRxiv (Cold Spring Harbor Laboratory) · 2024-09-15

  preprintOpen accessSenior author

  Abstract The “default mode” of cognition refers to the tendency to simulate internal experiences, rather than attending to external events in the moment. But in some contexts, external focus can become captivating enough to act as the default mode. To explore the relationship between prepotent internal and external default modes, we measured brain activity in forty participants using fMRI. Naturalistic movie clips were viewed, each one four times in sequence. When subjects were asked to focus attention on the videos, more mind-wandering events (distractions from the externally-focused task) occurred as the videos became less interesting with each repetition, and also when less engaging videos were presented. When subjects were asked to focus internally on breathing, more mind-wandering events (distractions from the internally-focused task) occurred when videos were most interesting (on the first repetition) and when more engaging videos were presented. In the fMRI data, inter-subject correlation, within-subject correlation, and GLM analyses found similar fronto-parietal networks engaged in transitions between default-controlled states regardless of the internal-external distinction, indicating more overlap in internal-external processing than previously assumed. We suggest that whether the default state is internal or external, and whether the sources that disrupt it are internal or external, depend on context.

  PublisherDOI

• The CIA and the American Religious Imagination

  2024-01-01

  other1st authorCorresponding

  During World War II and the Cold War, US intelligence officers studied and engaged religious traditions around the world in the service of US national security. This work was fueled by major political, cultural, and legal changes in US life at mid-century, including especially a new legal framework for national security institutions which empowered them with a broad mandate and little oversight. Spearheaded by the Office of Strategic Services (OSS) and carried forward by the Central Intelligence Agency (CIA), these efforts were called the “religious approach” to intelligence. Influenced by popular American ideas about the nature and function of religion as a global public good, US intelligence saw these religious traditions as an element of US national security. In this view, global religious systems were roughly similar in function and purpose even if they had superficial differences. This idea appealed to intelligence officers in part because it suggested that religious meaning was universally translatable across otherwise stark divides of language, culture, and ethnicity. These assumptions about the nature of religion were folded into an existing and powerful tradition of American exceptionalism, encouraging intelligence officers to view the United States and the world’s religions as natural allies. At the same time, these ideas about religion and its role in national security boomeranged back onto US religious groups, empowering some while threatening others viewed as insufficiently loyal to US national security. In incorporating this idea into their work, American intelligence officers projected their belief systems and certainties outward around the globe.

  PublisherDOI

• Unexpected Benefits of Self-Modeling in Neural Systems

  arXiv (Cornell University) · 2024-07-14

  preprintOpen accessSenior author

  Self-models have been a topic of great interest for decades in studies of human cognition and more recently in machine learning. Yet what benefits do self-models confer? Here we show that when artificial networks learn to predict their internal states as an auxiliary task, they change in a fundamental way. To better perform the self-model task, the network learns to make itself simpler, more regularized, more parameter-efficient, and therefore more amenable to being predictively modeled. To test the hypothesis of self-regularizing through self-modeling, we used a range of network architectures performing three classification tasks across two modalities. In all cases, adding self-modeling caused a significant reduction in network complexity. The reduction was observed in two ways. First, the distribution of weights was narrower when self-modeling was present. Second, a measure of network complexity, the real log canonical threshold (RLCT), was smaller when self-modeling was present. Not only were measures of complexity reduced, but the reduction became more pronounced as greater training weight was placed on the auxiliary task of self-modeling. These results strongly support the hypothesis that self-modeling is more than simply a network learning to predict itself. The learning has a restructuring effect, reducing complexity and increasing parameter efficiency. This self-regularization may help explain some of the benefits of self-models reported in recent machine learning literature, as well as the adaptive value of self-models to biological systems. In particular, these findings may shed light on the possible interaction between the ability to model oneself and the ability to be more easily modeled by others in a social or cooperative context.

  PublisherOA PDFDOI

Recent grants

• NIH Grant R01NS046407

  NIH · $1.1M · 2010

Frequent coauthors

• John Fea

  Hankuk University of Foreign Studies

  81 shared

• Ernest Hocking

  Washington University in St. Louis

  81 shared

• Michael Thompson

  International Institute for Applied Systems Analysis

  81 shared

• Matthew D. Sutton

  Morehead State University

  81 shared

• John D. Wilsey

  Southern Baptist Theological Seminary

  81 shared

• John Danforth

  Washington University in St. Louis

  81 shared

• Conroy-Krutz

  Washington University in St. Louis

  81 shared

• Amanda Porterfield

  81 shared