Threat as a feature in visual semantic object memory

Abstract

Threatening stimuli have been found to modulate visual processes related to perception and attention. The present functional magnetic resonance imaging (fMRI) study investigated whether threat modulates visual object recognition of man-made and naturally occurring categories of stimuli. Compared with nonthreatening pictures, threatening pictures of real items elicited larger fMRI BOLD signal changes in medial visual cortices extending inferiorly into the temporo-occipital (TO) "what" pathways. This region elicited greater signal changes for threatening items compared to nonthreatening from both the natural-occurring and man-made stimulus supraordinate categories, demonstrating a featural component to these visual processing areas. Two additional loci of signal changes within more lateral inferior TO areas (bilateral BA18 and 19 as well as the right ventral temporal lobe) were detected for a category-feature interaction, with stronger responses to man-made (category) threatening (feature) stimuli than to natural threats. The findings are discussed in terms of visual recognition of processing efficiently or rapidly groups of items that confer an advantage for survival.

Keywords: category; emotion; fMRI; feature; memory; object; semantic; threat; visual.

Figure 1

Sample visual stimuli for threatening, nonthreatening, “real” items', and a corresponding visual “scrambled” item.

Figure 2

Threat Feature Main Effect. Foci of signal change detected for the main effect of threat using whole‐brain 2 (Threatening vs. Nonthreatening) × 2 (Man‐Made vs. Natural) ANOVA. Color‐scaling is based on F‐values for the main effect, with the voxel‐wise F _minimum(1,17) = 10.4, P < 0.005, and the brain‐wise cluster k = 153, with P < 0.05. The direction of the difference (Threatening > Nonthreatening) is reported in more detail in the text. Data are displayed on the axial view of a template brain with the right side of the figure corresponding to the right side of the brain.

Figure 3

Threat Feature Main Effect from Peak Voxel. Illustration of the main effect of threat from the whole‐brain 2 (Threatening vs. Nonthreatening) × 2 (Man‐Made vs. Natural) ANOVA. Data are from the voxel (LPI coordinates = −7, −100, 0 mm) having the highest F‐value (F(1,17) = 25.47, P < 0.005) within the significant cluster. Errors bars represent SEM.

Figure 4

Category–Feature Interaction Effects. A. Foci of signal change detected using whole‐brain 2 (Threatening vs. Nonthreatening) × 2 (Man‐Made vs. Natural) ANOVA. Color‐scaling is based on F‐values for the interaction effect, with the voxel‐wise F _minimum(1,17) = 10.4, P < 0.005, and the brain‐wise cluster k = 153, with P < 0.05. B. Contrasts comparing man‐made to natural threatening stimuli within the interaction ROIs depicted in A. C. Contrasts comparing man‐made to natural nonthreatening stimuli within the interaction ROIs depicted in A. For B & C, color‐scaling is based on the t‐values for the contrasts, with red‐to‐yellow indicating higher percent signal change for man‐made versus natural and blue‐to‐cyan indicating higher percent signal change for natural compared with man‐made and with the voxel‐wise |t _minimum| = 1.7, P < 0.05. Data are displayed on the axial view of a template brain with the right side of the figure corresponding to the right side of the brain.

Figure 5

Category–Feature Interaction Effects. Illustration of the Category × Feature interaction effects from the whole‐brain 2 (threatening vs. nonthreatening) × 2 (Man‐Made vs. Natural) ANOVA. Data are from the voxels (LPI coordinates Left = −42, −90, −7 mm, and Right = 45, −62, −10 mm) having the highest F‐values (F _left(1,17) = 29.84, P < 0.005, and F _right(1,17) = 27.04, P < 0.005) within the significant cluster. Error bars represent SEM.