Object semantics is differently accessed from visual short-term memory by people suffering from Mild Cognitive Impairment

Objects semantically inconsistent with their scene context (e.g., a torch vs. a toothbrush in a bathroom) capture early overt attention and are better remembered. Yet, it remains unexplored whether such mechanisms of object-scene integration are still functioning in people suffering from Mild Cognitive Impairment (MCI). In two independent eye-tracking studies: (a) younger (N = 26) and healthy older participants (N = 24), and (b) people with MCI (N = 30) and healthy age-matched controls (N = 32) were asked to detect whether an object, either consistent or inconsistent with its embedding scene, changed on its identity (became another object), location (moved to another position) or concurrently both these features. Changes occurring on objects encoded as inconsistent were better detected than consistent ones by all groups. Oculomotor responses during successful detections reveal subtle differences between the MCIs and all other groups. MCIs were less likely to fixate the critical object as the first target when it changed in identity, and it was encoded as inconsistent (becoming consistent). Once fixated MCIs spent significantly longer on it when it changed only in identity. Like healthy participants, MCIs benefit from object semantics to successfully recall from visual short-term memory (VSTM). Nevertheless, they display slight differences in their early overt attention indicating that object semantics may not immediately be ready in VSTM, especially when the object is no longer predictable (i.e., it is consistent). Overall, predictive mechanisms of object-scene integration are still active in pathologically aged individuals, even though with interesting, possibly compensatory, changes.