Table 1.
Error Measures Used
Error name | Error calculation | Error description |
---|---|---|
Raw error | eRij = dIij – dAij | Response error for object i, turn j, relative to actual target direction (dA) on a circular scale (dI = indicated target direction) |
Linear error | eLij = eRij ± 360° | Raw error (eR) adjusted to be within 180° of a particular reference direction (e.g., 0°, eLij – 1) |
Heading error | a | Mean (using circular statistics) raw error (eR) for all 4 objects for a particular turn (j) |
Deviation error | eDij = eRij – eHj | Raw error (eR), for object i adjusted by mean eR for all 4 objects on corresponding turn (j) |
Adjusted linear error | eALij = eLij – eLij–1 | Linear error (eL) for object i on turn j adjusted by eL for object i on turn j – 1 |
Adjusted linear deviation error | Adjusted linear error (eAL) for object i adjusted by mean eAL for all objects on corresponding turn | |
Angular separation error | Absoluteb error in angular separation between a given pair of objects (i1, i2) on a given turn (j) | |
Position error | Absolute sum of adjusted linear error for a given pair of objects (i1, i2) on a given turn (j) |
If , then eHj = eHj + 180.
Absolute differences and sums were used so that the error scores could be aggregated within and across participants without positive and negative errors canceling each other out. Transforming the differences and sums in this way also rendered irrelevant the issue concerning which way around the circle angular separation was measured.