Top-down modulation of unconscious 'automatic' processes: A gating framework

Abstract

In classical theories of automaticity, automatic processes are usually thought to occur autonomously and independently of higher level top-down factors (e.g., Posner & Snyder, 1975). However, already Neumann (1984) pointed out that the cognitive system has to be configured in a certain way for automatic processes to occur. In extension of his work, I propose a gating framework to account for the influence of top-down factors such as attention, intention and task set on automatic processes such as masked response or semantic priming. It is assumed that task representations held in prefrontal cortex regulate the gain of neurons in visual and sematic association cortex thereby modulating the effects of unconsciously perceived masked stimuli on further 'automatic' information processing steps. In support of the postulated gating framework, recent studies demonstrated a top-down modulation of automatic processes. Behavioral and electrophysiological studies with the masked response priming and semantic priming paradigms show that masked priming effects crucially depend (i) on temporal attention to the masked prime, (ii) on intentions or action plans and (iii) on the task set active immediately before masked prime presentation. For instance, masked semantic priming was only observed when the preceding task set required the orientation to semantic word features, but not when it required orientation to perceptual word features. These results support the view that unconscious automatic processes are modulated by top-down factors. They are suggestive of a gating mechanism which orchestrates the conscious and unconscious information processing streams.

Keywords: automatic processes; masked response priming; masked semantic; priming; top-down control; unconscious perception.

Figure 1.

ERP priming effects. Absolute mean voltage difference between semantically unrelated and related word pairs (ERP priming effects) in the N400 time window at centro-parietal electrodes as a function of masking and prime-target SOA. Potentials were collapsed across hemispheres. This figure shows the qualitatively different time courses for unmasked and masked N400 priming effects (after Kiefer and Spitzer, 2000).

Figure 2.

Plots of (A) masked behavioral and (B) masked parietal N400 priming effects as a function of the sensitivity measure d’ in the masked visibility test. The plots also show the linear regression function (after Kiefer, 2002b).

Figure 3.

Outline of the gating framework. Task information (relevant stimulus dimensions, spatial and temporal stimulus information etc.) held in prefrontal areas modulates the gain of neurons in sensory areas through far reaching connections. Hereby, processing pathways in congruency with the represented task information are enhanced while other processing pathways are inhibited.

Figure 4.

Schematic depiction of sample congruent and incongruent trials (a) and response times for the three conditions (b) in Experiment 1 of the Naccache et al. (2002) study. The motor response was congruent when the prime and the target numbers were both either greater than 5 or less than 5; if one was greater than 5 and the other was less than 5, they were incongruent. Response priming effects were only obtained when the target was presented after a fixed time interval (after Naccache et al., 2002).

Figure 5.

Temporal sequence of one trial of the temporal cueing procedure. The masked prime word was presented either 200 ms or 800 ms following a cue, which prompted subjects to attend to the stimulation stream (after Kiefer and Brendel, 2006).

Figure 6.

Attentional modulation of ERP priming effects. Mean voltages from centro-parietal electrodes in the time window (A) of the descending N400 (200-399) and (B) of the N400 peak (400-599 ms) as a function the cue prime interval (CPI) and prime-target SOA (Experiment 1). Voltages were collapsed across electrode sites. In both time windows N400 priming effects were largest at the short CPI/short SOA condition demonstrating an attentional modulation of masked semantic priming (after Kiefer and Brendel, 2006).