关键词: SPT效应, 指令记忆, 多通道加工理论

Abstract: When learners encode a series of verbal instructions through action encoding (i.e., physically performing the described actions) rather than verbal encoding (i.e., merely listening to or reading the instructions), they exhibit significantly better recall performance, a phenomenon known as the subject-performed task (SPT) effect. Multimodal theory posits that this advantage arises because action encoding recruits multiple sensory channels beyond the verbal modality, including kinesthetic (motor execution), visual (e.g., observing objects and actions), and even tactile input (e.g., touching object surfaces), thereby enriching memory traces and enhancing their retrievability (Bäckman & Nilsson, 1984; Engelkamp & Zimmer, 1984). Building on this framework, the present study examined how the conventionality of action–object pairings in verbal instructions moderates the multimodal dependencies of the SPT effect, with the aim of distinguishing critical from auxiliary sensory contributions. Three experiments were designed, and all experiments used familiar everyday objects. In Experiment 1 (N = 200), we manipulated instruction conventionality (conventional, e.g., “sharpen a pencil”; non-conventional, e.g., “poke a pencil”) and the presence versus absence of object pictures (providing visual cues without physical interaction). Results showed that conventional instructions elicited a SPT effect (i.e., performance in the action encoding condition was superior to that in the verbal encoding condition) regardless of whether pictures were present. In contrast, non-conventional instructions failed to produce the SPT effect even when object pictures were available. We hypothesized that non-conventional pairings lack pre-existing action schemata, and static visual input alone is insufficient to support online construction of action–object bindings. To test this hypothesis, Experiment 2 (N = 58) replaced pictures with real, manipulable objects, enabling participants to physically interact with them during encoding. Under these conditions, non-conventional instructions yielded a significant SPT effect when objects were present but not when absent, indicating that sensorimotor interaction, not mere visibility, is essential for low-conventionality materials. However, because real objects concurrently engage multiple sensory channels, the paradigm in Experiment 2 cannot isolate the unique contributions of vision, touch, or other modalities. To disentangle these influences, Experiment 3 (N = 54) employed an orthogonal design that systematically manipulated visual access (blindfolded vs. unblindfolded) and tactile fidelity (wearing gloves vs. bare-handed) during object interaction, with a verbal-only baseline. Results revealed that the SPT effect emerged only when vision was available, irrespective of tactile condition. Specifically, recall performance was significantly higher in both “vision + touch” and “vision + no-touch” action conditions than in the verbal baseline (ps < .001), whereas “no-vision + touch” and “no-vision + no-touch” conditions did not differ from verbal encoding (ps > .999). These findings demonstrate that vision rather than touch is necessary for the SPT effect with non-conventional instructions. Collectively, these findings reveal a dynamic, experience-dependent architecture of multimodal contributions: for high-conventionality instructions, kinesthetic input alone suffices to drive the SPT effect via internalized action schemata; for low-conventionality instructions, visual input provides the structural scaffold for online action–object binding, kinesthesia supplies self-referential enactment, and tactile information plays no independent role. These findings challenge the traditional view of multimodal enhancement as a simple additive process and instead support a context-sensitive, dynamically weighted model in which sensory channel contributions are flexibly calibrated by the interplay between prior knowledge and current perceptual input. This advances multimodal memory theory and informs practical interventions for populations with sensory or neurodevelopmental differences (e.g., autism spectrum disorder), where optimizing visual–motor integration may enhance instruction memory without relying on tactile precision.

Key words: SPT effect, Instruction memory, Multimodal processing theory