The development of neuromorphic systems necessitates the use of memcapacitors that can adapt to optoelectronic modulation. Two-dimensional (2D) materials with atomically thin features and their derived heterostructures are able to allow for controlling local transfer of charge carrier but reports on 2D materials-enabled capacitive-type photoelectric synapses have not been experimentally exploited yet. Herein, MXene-TiO2 heterostructured iontronic neural devices based on ion-dynamic capacitance enabling optoelectronic modulation are designed. According to the electrochemical insight, under UV light illustration, photoexcited electrons in TiO2 flow to MXene, leading to the localized accumulation of electrons as the trapping center and thus inducing the embedding of H+ for participating in the pseudo-intercalation. On removing the UV light, a part of trapped H+ are not instantly returned to the initial state. As a result, this memcapacitor features hysteresis ion-dynamic capacitance under optoelectronic modulation. Through assessing its applicability to neuromorphic computing, this memcapacitor achieves the high recognition accuracy (93.5%) of handwritten digits by recognizing and sharpening the input signal trajectory.