Now you see it,now you don't: The context dependent nature of category-effects in visual object recognition

In two experiments, we test predictions regarding processing advantages/disadvantages for natural objects and artefacts in visual object recognition. Varying three important parameters—degree of perceptual differentiation, stimulus format, and stimulus exposure duration—we show how different category-effects can be provoked in normal subjects on the same task. We interpret the results in light of the Pre-semantic Account of Category Effects (PACE; Gerlach, 2009), and conclude that category-effects do not reflect absolute processing differences between categories. Rather, category-effects are products of common operations which are differentially affected by the structural similarity among objects (with natural objects being more structurally similar than artefacts). The potentially most important aspect of the present study is the demonstration that category-effects are very context dependent; an aspect which has been neglected in earlier literature on category-specificity. PACE provides a useful framework in this respect, as it specifies how category-effects are influenced by changes in task and stimulus characteristics.     

The role of attention in the contextual enhancement of visual memory for natural scenes

The role of attention in memory for objects in natural scenes was investigated using a visual memory. In Experiment 1, participants were asked to memorize six cued (to be attended) objects in a natural scene, and were subsequently tested on one of the cued objects. Four types of test scene images were created by jumbling different sections in scene's background: Attended sections changed, unattended sections changed, both sections changed, and both unchanged. In Experiment 2, the procedure was the same as that of the jumble condition except that scenes to be memorized were also jumbled. Results showed that jumbling of attended sections reduced memory performance, whereas jumbling unattended sections did not, irrespective of the regularity of scene to be memorized. This finding suggests that attention plays an important role in mental construction of a natural scene representation, and leads to enhanced visual memory.     

Additive effects of inhibiting attention to objects and locations in three-dimensional displays

One of the processes thought to underlie visual selection works by biasing attention away from either recently examined locations or objects. The extent of this “inhibition” is greatest when the inhibited object and the inhibited location coincide. In Experiment 1, rectangles are presented stereoscopically at different depths but at similar positions horizontally and vertically. Here, any inhibition should be based solely on a spatial code, as the objects, the rectangles are clearly separate objects. In Experiment 2, the corners of the rectangles are joined to produce a single cuboid that extends in depth space. Now inhibition based on both spatial and object codes should be seen because even when on different depth planes the cue and target are associated with the same object. Consistent with our understanding of the additive effects of inhibition of space and object codes, the extent of inhibition in the second study is almost double that of the first. The results further suggest that space-based inhibition operates within a two-dimensional representation while object-based inhibition utilizes a three-dimensional representation.     

Eccentricity biases of object categories are evident in visual working memory

In high-order object areas, face-selective areas prefer centrally presented stimuli, whereas building-selective areas prefer peripherally presented stimuli (Levy et al., 2001). We investigated whether this eccentricity bias was also evident in visual working memory. In Experiment 1, we found that working memory performance for faces decreased towards the periphery while the performance for buildings remained unchanged across different eccentricities. To rule out the possibility that lower level features influence these results, we manipulated the spatial frequency of faces and buildings (Experiment 2) and the spatial layout information of the buildings (Experiment 3). In both of the experiments, we replicated the results of Experiment 1, even when these lower level features of stimuli were controlled. Consistent with previous findings, the current results suggest that each object category is processed in a different manner depending on the eccentricity. This eccentricity bias is likely the result of how the high-order object areas represent different object categories.     

Why does vantage point affect boundary extension?

To determine if layout affects boundary extension (BE; false memory beyond view boundaries; Intraub & Richardson, 1989), 12 single-object scenes were photographed from three vantage points: central (0°), shifted rightward (45°), and shifted leftward (45°). Size and position of main objects were held constant. Pictures were presented for 15 s each and were repeated at test with their boundaries: (a) displaced inward or outward (Experiment 1: N=120), or (b) identical to the stimulus views (Experiment 2: N=72). When participants adjusted test boundaries to match memory, BE always occurred, but tended to be smaller for 45° views. We propose this reflects the fact that more of the 3-D scene is visible in the 45° views. This suggests that scene representation reflects the 3-D world conveyed by the global characteristics of layout, rather than the 2-D distance between the main object and the boundaries of a picture.     

Masked partial priming of letter perception

In three experiments testing 178 subjects, letter targets were preceded by briefly presented, pattern-masked primes formed by deleting pixels in a larger or smaller version of the target stimulus. In Experiment 1A (alphabetic decision)and Experiment 1B (letter naming) a slight advantage was observed for global primes (alternate pixels deleted) compared with junction primes (midsegment information removed). This advantage was stronger at 50 ms prime exposures than at 30 ms exposures in the naming task. In Experiment 2 (letter naming), midsegment primes (with junction information removed) produced faster latencies than did junction primes. This result was replicated in a third experiment and was shown to be independent of target letter case and the relative size of prime and target stimuli. The same midsegment and junction primes did not facilitate performance compared to neutral primes in the alphabetic decision task. These results suggest that masked partial priming of letter naming can be usefully applied to the study of basic processes in letter perception.     

Visual search for expressions and anti-expressions

We used visual search to explore whether the preattentive mechanisms that enable rapid detection of facial expressions are driven by visual information from the displacement of features in expressions, or other factors such as affect. We measured search slopes for luminance and contrast equated images of facial expressions and anti-expressions of six emotions (anger, fear, disgust, surprise, happiness, and sadness). Anti-expressions have an equivalent magnitude of facial feature displacements to their corresponding expressions, but different affective content. There was a strong correlation between these search slopes and the magnitude of feature displacements in expressions and anti-expressions, indicating feature displacement had an effect on search performance. There were significant differences between search slopes for expressions and anti-expressions of happiness, sadness, anger, and surprise, which could not be explained in terms of feature differences, suggesting preattentive mechanisms were sensitive to other factors. A categorization task confirmed that the affective content of expressions and anti-expressions of each of these emotions were different, suggesting signals of affect might well have been influencing attention and search performance. Our results support a picture in which preattentive mechanisms may be driven by factors at a number of levels, including affect and the magnitude of feature displacement. We note that indirect effects of feature displacement, such as changes in local contrast, may well affect preattentive processing. These are most likely to be nonlinearly related to feature displacement and are, we argue, an important consideration for any study using images of expression to explore how affect guides attention. We also note that indirect effects of feature displacement (for example, changes in local contrast) may well affect preattentive processing. We argue that such effects are an important consideration for any study using images of expression to explore how affect guides attention.     

Changes in boundary extension effect during spatial confinement

Changes in perception during space missions are usually attributed to microgravity. However, additional factors, such as spatial confinement, may contribute to changes in perception. We tested changes in scene perception using a boundary extension (BE) paradigm during a 105-day Earth-based space-simulation study. In addition to the close-up/wide-angle views used in BE, we presented two types of scenes based on the distance from the observer (proximal/distant scenes). In crew members (n = 6), we found that BE partly increased over time, but the size of BE error did not change in the control group (n = 22). We propose that this effect is caused by an increasing BE effect in stimuli that depict distant scenes and is related to spatial confinement. The results might be important for other situations of spatial confinement with restricted visual depth (e.g., submarine crew, patients confined to a bed). Generally, we found a larger BE effect in proximal scenes compared with the distant scenes. We also demonstrated that with no feedback, subjects preserve the level of the BE effect during repeated measurements.     

Dimension‐specific intertrial facilitation in visual search for pop‐out targets: Evidence for a top‐down modulable visual short‐term memory effect

In visual search for pop‐out targets, target detection is expedited when the target‐defining dimension (e.g., colour) is repeated on consecutive trials, but not (or less) when the target feature (e.g., red) is repeated (Found & Müller, 1996). The present study examined whether this dimension‐specific intertrial facilitation (ITF) is mediated by an explicit or an implicit memory of the target on the previous trial. In Experiment 1, observers had no above‐chance knowledge of the target dimension (or feature) when this information was not to be encoded and retained explicitly; yet, they exhibited dimension‐specific ITF. In Experiment 2, the nature of the ITF effect was unaffected by whatever attribute of the target was to be explicitly encoded and memorized, its dimension or feature; however, ITF was increased compared to a no‐memory control condition. In Experiment 3, ITF was not increased when observers had to determine the target dimension, but without having to maintain a memory of it during the intertrial interval. These results suggest that, although an implicit memory is sufficient for producing dimension‐specific ITF, ITF is top‐down modulable by explicit memory instructions.     

Dynamic states in working memory modulate guidance of visual attention: Evidence from an n-back paradigm

Items held in working memory (WM) can automatically bias attention when they reappear in visual displays. Recent evidence, however, suggests that WM biases of attention may be reduced under certain conditions, for example with increasing memory load. We employed a dual task paradigm to investigate how WM biases are affected by dynamic updating of memory contents. 1-back and 2-back versions of a memory task with colour stimuli were interrupted at intervals by an unrelated visual search task. Reappearance in the search display of the item that was currently active in WM guided attention, while suppressed or inactive items did not. We conclude that the rapid updating of memory contents facilitates the shifting of memory representations into different activity states on a moment-to-moment basis. The finding is consistent with models that propose that only one item can be “active” in WM at any one time to guide attention.