Semantic influences on object detection: Drift diffusion modeling provides insights regarding mechanism
Fig 7
Potential mechanisms operative in Skocypec and Peterson’s object detection task [19].
Solid black lines with double-headed arrow endings indicate reentrant activity initiated by the object in the bipartite test display, both within the semantic network and between the semantic network and a lower-level representation of the test display (shown below the semantic network). Dashed lines with double-headed arrow endings indicate reentrant activation in the semantic network initiated by labels shown before the test displays, blue for valid labels, red or purple for invalid labels. Blue circles in the semantic network indicate semantic representations of the object in the test display in the various portions of the semantic network (e.g., context, object, etc.). Red & purple circles indicate semantic representations of objects denoted by invalid labels. Object NP = neural population representing an object. A) control labels-absent study. B) Valid labels in labels-present studies. (C-D) Invalid labels in labels-present studies; C) different superordinate-level invalid label as in study 1. D) same superordinate-level invalid label as in study 2. The green arrow labeled “R” emerging from the object NP in the semantic network indicates the participant’s right or left response; it’s shown in black to indicate that although the drift rate and threshold are affected by activity throughout the semantic network which affects activity within the NP representing the object, the R/L response must be dominated by recurrent activity between that NP and a lower-level representation of the object in the display.