Rhinal Cortex Enables Acquisition of Trace Conditioning In A Dopamine-Dependent Manner

Abstract

The rhinal cortex (RCx), comprising perirhinal, ectorhinal, and entorhinal cortices, is a major gateway between higher-order sensory association networks and the hippocampal memory system and has been implicated broadly in recognition memory, associative representations, and stimulus–outcome learning. However, how RCx contributes to acquiring cue–reward associations across temporal gaps remains incompletely understood. Given its multisensory integration and strong connectivity with memory and neuromodulatory circuits, we hypothesized that RCx plays a critical role in trace conditioning, in which a temporal gap separates cue and reward and requires a neural mechanism to bridge the interval.

We trained head-fixed, water-restricted mice on an auditory trace conditioning task in which a brief tone was followed by a water reward after a 1-s trace interval. Using a data-driven behavioral framework, we classified training sessions into distinct learning phases based on unsupervised clustering of licking dynamics. Importantly, this licking-defined phase structure was independently supported by learning-related changes in pupil size, consistent with progressive modulation of arousal state across training. Local inactivation of RCx using muscimol or chemogenetic inhibition impaired acquisition of cue–reward associations but did not disrupt the expression of learned associations in expert animals, indicating that RCx is essential for learning but dispensable once the association is established.

Fiber photometry recordings from excitatory RCx neurons revealed a cue-locked calcium response that progressively strengthened and became more temporally extended with learning, consistent with the emergence of a sustained cue-related activity spanning the trace interval. To dissociate task-evoked neural dynamics from concurrently measured behavioral variables, we implemented a generalized linear model that incorporated task events together with multidimensional behavioral readouts. Across model variants, the learning-dependent enhancement and temporal extension of cue-locked activity remained robust, supporting the conclusion that strengthened cue encoding in RCx is a dominant signature of acquisition. Consistent with this interpretation, neural decoding analyses showed that RCx-derived features predicted both discrete learning phases and session-level behavioral performance, with cue-related features carrying stronger predictive information than reward-related features.

Complementary recordings using a genetically encoded dopamine sensor demonstrated that dopamine release in RCx, initially confined to reward delivery, developed a prominent ramp from cue onset through the trace interval and declined at the expected reward time when reward was omitted. Furthermore, local infusion of the D1 receptor antagonist SCH23390 into RCx blocked acquisition of the trace association, establishing a causal role for D1 receptor activation in learning.

Together, these findings demonstrate that RCx supports trace conditioning by strengthening predictive cue representations across the trace interval, with dopamine-dependent D1R signaling contributing critically to the acquisition of this cue-dominant representation.

Date of Award	6 May 2026
Original language	English
Awarding Institution	City University of Hong Kong
Supervisor	Jufang HE (Supervisor) & Xi CHEN (Co-supervisor)