Self-motion can be perceived via podokinetic information, that is, based upon the movements of the legs during legged locomotion. This information can be integrated in order to perceive a path of travel through the environment (i.e., via podokinetic path integration). Two types of podokinetic information have been distinguished by analyzing the patterns of bias that result from manipulating the gait patterns used in direct-route homing tasks. Each type of podokinetic information has been associated specific groupings of gaits that support equivalent perceptual measurements of self-motion. Specifically, gaits are grouped if they can be varied across the outbound and inbound phases of a homing task (e.g., walking outbound and jogging inbound) and the accuracy of homing task performances does not differ from matched gait control conditions.Recently, it was theorized that different types of podokinetic information are related to the differences in the kinematic form of limb motions in these groupings of gaits. Here we test an alternative hypothesis, namely, that attention plays a role in selecting the type of podokinetic information. In three experiments we manipulated the crawling gait patterns used in direct-route homing tasks. Consistent with our hypotheses, we observe that selfmotion is equivalently measured via crawling movement patterns that 1) have distinct kinematic forms, but that similarly direct participants' attention onto controlling the swing phase trajectories of their arms, and 2) have distinct inter-limb coordination patterns (i.e., pace vs. trot), but do not require attention to be specifically focused upon swing phase arm trajectories.