Five experiments are reported that were designed to examine learning, adaptation, and transfer across the actual 1-g environment, the vi1tual 1-g environment, and the vi1tual 0-g environment. The study aimed to, (1) validate the notion that virtual task environments provide a one-to-one mapping of actual task environments, whereby the intrinsic dynamic constraints on action are perceived to be invariant, (2) provide empirical evidence in support of virtual zero-gas an effective training environment for pre-adapting humans to the perception-action cycles of space, and (3) provide empirical evidence in support of virtual 1-g as an effective training environment for re-adapting humans to the perception-action cycles of Earth. While the results obtained from the five experiments clearly support these aims, with participants consistently demonstrating improvement or deterioration in te1ms of the task criterion (a decreased in absolute e1ror when tested in the same gravitational environment and an increase in absolute error when transferred across the two gravitational environments). The measure of error tendency (sign e1rnr) reflected inconsistencies in relation to the perception-action search strategies used within, and across, the three task environments. Consequently, three exploratory search strategies were identified (a negative-up, a positive-down, and an oscillating strategy) that exhibit an asymmetric relationship between the actual 1-g environment and both virtual environments, whereby participants employ the oscillating strategy more frequently during virtual to virtual environment transfer and the negative-up and positive down strategies during actual to virtual or virtual to actual environment transfer. The value of VR for the practice of zero-g perception-action cycles, as well as perception-action cycles in general, was highlighted and the nature (local and global) and impact of the three different search strategies on learning, adaptation, and transfer were discussed in relation to previous research.