Fig 1.
Depiction of experimental setup.
(A) Schematic depiction of two participants (re-enacted), a trained and an untrained one, interacting by looking, pointing, and talking while referencing objects of interest to come to a joint decision. On the cardboard stands in the foreground, participants’ individual notes are placed close to eye-height to mitigate data loss. (B) Depiction of the right participant’s eye tracker video feed showing a red dot indicating the gaze position as well as the left participant’s pointing hand.
Fig 2.
Example of a recurrence plot constructed from nominal time series data of two interacting participants’ gaze towards the 9 task relevant items. The boxes along the diagonal line from the bottom left to the top right indicate moments of prolonged joint attention. Unique for this example dyad, symmetrical converging structures on the top left and bottom right, mirrored across line of coincidence indicate that both participants revisited all items during approximately the last minute of the conversation.
Fig 3.
Average Diagonal Cross Recurrence Profiles (DCRPs) of gaze behavior between participants.
(A) Average DCRP of trained participants looking and untrained participants looking. Ribbons represent standard error. Recurrence on positive time lags indicates that the trained participant’s behavior (looking) leads the untrained participant’s behavior (looking) in time. (B) Average DCRP from the same data, now split by participant’s conversational turn.
Fig 4.
Average Diagonal Cross Recurrence Profiles (DCRPs) of pointing and gaze behavior between participants.
DCRPs of participant’s pointing and their respective counterpart’s looking, split by training level. Ribbons represent standard error. Recurrence on positive time lags indicate that the trained participant’s behavior (looking or pointing) is leading in time.
Fig 5.
Average Diagonal Cross Recurrence Profiles (DCRPs) of pointing and gaze behavior within each participant.
DCRPs represent participants’ pointing and their own looking, split by training level. Ribbons represent standard error. Note that this plot is constructed differently to the earlier ones as the DCRPs depict coupling of behavior within the same participants rather than across a dyad. Recurrence on positive time lags here indicates that pointing leads looking, whereas those on negative time lags indicate that looking leads pointing.