The contribution of context to visual object recognition

Looking to recognise: the pre-eminence of semantic over sensorimotor processing in human tool use

Alongside language and bipedal locomotion, tool use is a characterizing activity of human beings. Current theories in the field embrace two contrasting approaches: "manipulation-based" theories, which are anchored in the embodied-cognition view, explain tool use as deriving from past sensorimotor experiences, whereas "reasoning-based" theories suggest that people reason about object properties to solve everyday-life problems. Here, we present results from two eye-tracking experiments in which we manipulated the visuo-perceptual context (thematically consistent vs. inconsistent object-tool pairs) and the goal of the task (free observation or looking to recognise). We found that participants exhibited reversed tools' visual-exploration patterns, focusing on the tool's manipulation area under thematically consistent conditions and on its functional area under thematically inconsistent conditions. Crucially, looking at the tools with the aim of recognising them produced longer fixations on the tools' functional areas irrespective of thematic consistency. In addition, tools (but not objects) were recognised faster in the thematically consistent conditions. These results strongly support reasoning-based theories of tool use, as they indicate that people primarily process semantic rather than sensorimotor information to interact with the environment in an agent's consistent-with-goal way. Such a pre-eminence of semantic processing challenges the mainstream embodied-cognition view of human tool use.

Heightened Responses of the Parahippocampal and Retrosplenial Cortices during Contextualized Recognition of Congruent Objects

Context sometimes helps make objects more recognizable. Previous studies using functional magnetic resonance imaging (fMRI) have examined regional neural activity when objects have strong or weak associations with their contexts. Such studies have demonstrated that activity in the parahippocampal cortex (PHC) generally corresponds with strong associations between objects and their spatial contexts while retrosplenial cortex (RSC) activity is linked with episodic memory. However these studies investigated objects viewed in associated contexts, but the direct influence of scene on the perception of visual objects has not been widely investigated. We hypothesized that the PHC and RSC may only be engaged for congruent contexts in which the object could typically be found but not for neutral contexts. While in an fMRI scanner, 15 participants rated the recognizability of 152 photographic images of objects, presented within congruent and incongruent contexts. Regions of interest were created to examine PHC and RSC activity using a hypothesis-driven approach. Exploratory analyses were also performed to identify other regional activity. In line with previous studies, PHC and RSC activity emerged when objects were viewed in congruent contexts. Activity in the RSC, inferior parietal lobe (IPL) and fusiform gyrus also emerged. These findings indicate that different brain regions are employed when objects are meaningfully contextualized.

Ultrafast scene detection and recognition with limited visual information

Humans can detect target color pictures of scenes depicting concepts like picnic or harbor in sequences of six or twelve pictures presented as briefly as 13 ms, even when the target is named after the sequence (Potter, Wyble, Hagmann, & McCourt, 2014). Such rapid detection suggests that feedforward processing alone enabled detection without recurrent cortical feedback. There is debate about whether coarse, global, low spatial frequencies (LSFs) provide predictive information to high cortical levels through the rapid magnocellular (M) projection of the visual path, enabling top-down prediction of possible object identities. To test the "Fast M" hypothesis, we compared detection of a named target across five stimulus conditions: unaltered color, blurred color, grayscale, thresholded monochrome, and LSF pictures. The pictures were presented for 13-80 ms in six-picture rapid serial visual presentation (RSVP) sequences. Blurred, monochrome, and LSF pictures were detected less accurately than normal color or grayscale pictures. When the target was named before the sequence, all picture types except LSF resulted in above-chance detection at all durations. Crucially, when the name was given only after the sequence, performance dropped and the monochrome and LSF pictures (but not the blurred pictures) were at or near chance. Thus, without advance information, monochrome and LSF pictures were rarely understood. The results offer only limited support for the Fast M hypothesis, suggesting instead that feedforward processing is able to activate conceptual representations without complementary reentrant processing.

Biologically inspired scene context for object detection using a single instance

This paper presents a novel object detection method using a single instance from the object category. Our method uses biologically inspired global scene context criteria to check whether every individual location of the image can be naturally replaced by the query instance, which indicates whether there is a similar object at this location. Different from the traditional detection methods that only look at individual locations for the desired objects, our method evaluates the consistency of the entire scene. It is therefore robust to large intra-class variations, occlusions, a minor variety of poses, low-revolution conditions, background clutter etc., and there is no off-line training. The experimental results on four datasets and two video sequences clearly show the superior robustness of the proposed method, suggesting that global scene context is important for visual detection/localization.