Aliyar Ozercan
PhD Philosophy, University of Connecticut
Philosophy of Cognitive Science, Philosophy of Mind, Philosophy of Psychology,
Cognitive Developmental Psychology, Neurobiology of Language
Overview
I work at the intersection of philosophy, cognitive science, and psychology, using developmental evidence, computational modeling, and formal methods to clarify how social cognition emerges and how we should theorize its building blocks. A central focus of my research is infancy and early childhood as an empirical window into the architecture of mindreading, emotion understanding, and goal-directed social interaction. I develop conceptually precise accounts that connect philosophical questions to testable predictions, drawing on behavioral paradigms, decision-theoretic and logical tools, and neurocognitive perspectives. Alongside research on early social cognition and inquiry, I have a strong background in teaching across philosophy and psychology and experience in neuroimaging research and analysis.
Teaching
UConn Provost's Teaching Excellence Award
PI for 16 sections; TA for 16 sections
Average Evaluation: 4.61/5 (+.55 above inst. averages)
Current Research
Emotion-First Mindreading in Infancy: A Theory of Emotion
R&R at Mind and Language
-
Core claim: The earliest domain-specific mindreading capacity is a Theory of Emotion, where infants treat others’ emotions as structured appraisals rather than mere cues, enabling action-guiding predictions for social coordination.
-
Developmental ladder: A staged progression from resonance/matching, to category-level sensitivity, to direct representation that supports early, practical forms of metarepresentation in everyday interactions.
-
Empirical payoffs: Integrates evidence (emotion categorization, comforting/prosociality, humor, expectancy-violation), contrasts emotion-first with belief-first/desire-first views, and proposes testable behavioral and neural signatures plus tasks for longitudinal and cross-cultural adjudication.
The Puzzle of Early Deception: Lying Before Belief Representation
R&R at Cognition
-
Core claim: Infant pointing is not merely a learned gesture or a full-blown expression of propositional mindreading. It is an early form of social cognition in which infants regulate another agent’s attention relative to a practical or informational norm.
-
Architecture: The paper models pointing as adaptive social regulation, built from five components: a regulated variable, a target norm, an action repertoire, an implicit functional model, and an error detector. This explains why infant pointing is partner-sensitive, context-sensitive, and repair-capable.
-
Empirical payoffs: The account reframes the imperative/declarative distinction as a difference in target norms rather than two separate capacities, explains why lean conditioning accounts under-ascribe and rich mindreading accounts over-ascribe, and offers a minimal but structured lower bound for social cognition.
Published at Synthese
-
Core idea: Introduces Indefinite Targeting as a basic format for inquiry, an open template like "an F that meets condition C" that does not assume existence or uniqueness, allowing inquiry to begin, update with evidence, and remain the same inquiry through revision.
-
Norms for inquiry: Develops continuity constraints for when revisions preserve topic and stopping rules for when inquiry counts as complete, including cases with empty, multiple, or misdescribed targets (e.g., principled negative conclusions or plural answers).
-
Explanatory and empirical payoff: Uses classic cases (Neptune, Vulcan, two-thieves) to generate operational predictions about how search should change under conflict, how constraints enter measurement, and how outcomes should be scored, while contrasting with definite-description baselines and proposing tasks for adjudication across historical, formal, and experimental approaches.
Helping Without Beliefs: Teleological Alignment and the Minimal Architecture of Early Instrumental Helping
Under Review
-
Core claim: Toddlers' instrumental helping (selective, targeted, efficient) can be generated by a belief-free coordination system, Teleological Alignment, that tracks public goal structure and objective constraints, without presupposing belief attribution, shared commitment, or moral concern.
-
Mechanism: A predictive control loop with a Teleological Forward Model (anticipates efficient goal trajectories), a Disruption Monitor (treats breakdown as prediction error), and a Completion Release Mechanism (intervention as error-minimizing repair).
-
Payoff and predictions: Reinterprets classic helping evidence as outputs of fast teleological prediction, and predicts signature limits when public structure underdetermines a coherent completion (for example, epistemic opacity or constraint-incompatible attempts).
Reference Alignment and Repair as Optimal Control
Under Review
-
Core idea: The paper argues that reference alignment is not just about inferring what a speaker meant, but about choosing actions under residual uncertainty (e.g., when to wait, ask, commit, or tolerate ambiguity) in real time.
-
Model: It formalizes the listener as maintaining a belief distribution over candidate targets and selecting among Wait, Ground, Guess, and LetGo to maximize expected utility given delay and interruption costs, treating alignment as cost-sensitive control under uncertainty.
-
Predictions and payoff: An acceptance threshold tau gates when public commitment is rational, and the POMDP framing yields testable predictions about clarification frequency/latency as functions of stakes, costs, cue informativeness, and cognitive load (with worked cases for proper names, lexical ambiguity, and stance/irony).
Proto-Social Cognition: A Relational Framework for Infant Sociality
Under Review
-
Core puzzle and proposal: Explains the Participant–Spectator Paradox (infants as strong second-person interactants but weaker third-person predictors) by positing Proto-Social Cognition (PSC) as a distinct, early system that prioritizes interactional coordination over veridical prediction.
-
Architecture: PSC uses a relation-first, non-metarepresentational code r=<A, rho, T, theta> (agent, relation, target, graded access) plus a real-time control policy that regulates engagement via Maintain / Repair / Withdraw dynamics.
-
Payoff and predictions: Accounts for early social signatures (contingency, cross-modal integration, partner-general generalization) while predicting failures in tasks that constitutively require false-belief-style content embedding, and it offers discriminative, falsifiable predictions (for example, a coordinative-processing advantage when goal info is held constant). 
Emotion Reading as Mindreading in Infancy: A Complexity-Cost Criterion
Under Review
-
Problem: Infant social-cognition findings are often underdetermined because single measures (like looking time) can fit both “rich” mindreading interpretations and "lean" cue-based alternatives.
-
Proposal: Introduces a Complexity–Cost Criterion: attribute emotion as an agent-indexed hidden cause only when lean rivals must add ad hoc latent structure to match a conjunctive portfolio of signatures (not one effect).
-
Why emotion and how to test: Argues emotion is an especially strong testbed (elicitor specificity, resolution dynamics, cross-modal affordances) and sketches three experimental templates: Convergent Comforting, Causal-Specific Prediction, and Active Emotional Probing.
Beyond Belief: Dennett’s Standard for Mindreading Revisited
Under Review
-
Core claim: Challenges the false-belief "belief-benchmark" as the decisive test for Theory of Mind, arguing it rests on a misreading of Dennett (1978). Instead, it defends a content-agnostic criterion reading: what matters is whether attributing any mental state is the most parsimonious explanation of behavior.
-
Framework: Makes Dennett’s key clause (explanatory economy) empirically tractable by formal model comparison between a State-Attribution model and two deflationary rivals: Stimulus–Response and Teleological models.
-
Payoff and program: Applies the framework to early “precursor” domains (comforting, desire/intention tracking, knowledge tracking), arguing existing studies often rule out only the simplest competitors, and proposing task families to generate decisive evidence and unify early and later mindreading as gradual enrichment of representable state-space. 
Domain-Specific Sub-Theories of Mind and Their Developmental Assembly
Under Review
-
Core claim: Argues that the false-belief bottleneck has distorted the study of early Theory of Mind. False-belief reasoning is a stringent test of mindreading, but not its only form. Earlier forms of social understanding already count as genuine mindreading when they support structured prediction of what others feel, want, know, or are trying to do.
-
Framework: Develops a Sub-Theory of Mind architecture in which mindreading is assembled from partially distinct components for motion, emotion, intention/desire, knowledge, and belief. Earlier components support genuine but more basic forms of social prediction, while belief introduces the more demanding capacity to represent reality-conflicting mental states.
-
Payoff and program: Reframes the rich-versus-lean debate by showing that different infant tasks recruit different components rather than a single all-or-nothing capacity, yielding a clearer research program for developmental, comparative, and atypical social cognition.
Prepositional Attitudes: Fear, Anger, and Desire Between Propositionalism and Objectualism
In preparation
-
Core claim: Argues that some attitudes, especially fear of, anger at, and desire for, are constitutively non-propositional and are best understood as target-plus-mode states.
-
Develops a middle position between propositionalism and objectualism, showing why such attitudes resist reduction both to hidden propositions and to bare object-directedness.
From the last conference I organized, Millikanfest(left to right) Dennett, Bill (my advisor) and me