Objects in daily life are not always immediately recognized or known to the viewer, but they generally are easier to identify when the contextual information surrounding them is taken into account. The influence of contextual information depends on how strong the relationship is between object and context, and closer relationships increase the likelihood that the context will contribute to the recognition of the objects (Davenport and Potter, 2004). Functional magnetic resonance imaging (fMRI) studies examining relationships between object and context report that objects having strong associations with specific scene contexts (e.g., a tennis ball with a tennis court) elicit different brain activity than objects with weak or nonspecific associations (e.g., a generic rubber ball is associated with sports more broadly; Bar and Aminoff, 2003; Bar et al., 2005; Aminoff et al., 2007; Kim and Biederman, 2011).

Such studies have identified two cortical areas that together form parts of a larger neural network and play complementary roles in processing contextual associations between scenes and objects: the parahippocampal (PHC) and the retrosplenial (RSC) cortices. Higher activity in the PHC is mostly observed when objects with strong scene context associations are viewed; accordingly, the PHC appears to process place-related contextual information for objects with strong associations (Bar and Aminoff, 2003; Gronau et al., 2008; Kveraga et al., 2010). Furthermore, one study shows that even non-spatial scene associations congruently engage the PHC, which may suggest that the PHC is engaged with context processing more generally (Aminoff et al., 2007). A few studies have reported that the RSC also shows heighted activity for objects with strong associations and point to broader neuroscience literature linking RSC activity with episodic memory; thus, theorizing that the RSC accesses past memories when evaluating contextual associations (Bar and Aminoff, 2003; Vann et al., 2009). Numerous animal studies have investigated RSC activity during context processing (Keene and Bucci, 2008, 2009; Robinson et al., 2011). However, the relationships between RSC activity and context have not been directly examined in humans.

The interpretive influence of scene contexts on the recognition of objects does not only depend on the strength of their relationship, but it also depends on whether contexts have congruent associations with the object and how contextual information is applied to object recognition. Congruent context-object relationships refer to objects that typically appear in a given scene, whereas incongruent relationships refer to scenes in which the object appears odd or inappropriate. Several studies have shown that objects presented in congruent scenes are easier to recognize than objects placed in incongruent scenes or no scene at all (Oliva and Torralba, 2007). The context is more critical when the target object is difficult to recognize on its own (Klink et al., 2012; Dyck and Brodeur, 2015), such as when an object is unfamiliar, viewed from an unusual point of view, or when visibility is compromised by light and distance. In these circumstances, objects are visually ambiguous and should activate brain regions mediating the object-scene relationship, namely the PHC and RSC, when they are seen in a congruent scene (Rémy et al., 2014). While previous research has established that the PHC and RSC are activated by the object-scene associations that are initiated during perception of focal objects, which are centrally located in the visual field and presented in isolation, they have not generally examined whether these regions are involved when object recognition is viewed within a scene context.

The influence of scene context on object recognition brain processes has been mostly studied with event-related potentials (ERPs). These studies have shown that the electrophysiological brain activities of object recognition depend on how well the objects are congruent with the scene in which they appear (Ganis and Kutas, 2003; Mudrik et al., 2010, 2014; Demiral et al., 2012; Võ and Wolfe, 2013). Objects presented in incongruent scenes resemble the N400 response, which has been previously associated with incongruity, and elicit N390 and other negative ERPs over the central-frontal electrodes. To control for incongruity, a recent study from our lab investigated the effects of congruent contexts on ambiguous and unambiguous objects compared to a neutral condition in which presented objects neither appeared well-suited nor ill-suited (Dyck and Brodeur, 2015). The beneficial influence of congruent scenes was found on electrodes covering fronto-central and posterior areas. This study also demonstrated that ERPs of ambiguous objects viewed in congruent contexts resemble ERPs of unambiguous objects, meaning that the context effect mostly operates by reducing the ambiguity of percepts.

To our knowledge, Kirk (2008) has conducted the only fMRI study that directly investigated the influence of scene on the perception of visual objects. Their subjects made aesthetic judgments of stimuli including object in congruent scenes and objects in incongruent scenes. Greater activities for congruent stimuli were found in the lateral occipital cortex, the inferior parietal lobule, and the PHC. When the relationship between the object and the scene was unexpected, the frontal and parietal cortices, in addition to the RSC, were activated. In a subsequent study, Kirk et al. (2009) reported that frontal activity was related to a bias of judgment induced by the context. While these results show that the PHC and RSC activity are influenced by object-scene congruity, they showed that other brain areas, including the frontal and the parietal areas, are involved as well.

Building on previous research from our group, a study was conducted to evaluate brain activities during correct recognition of objects appearing in incongruent and neutral scenes. In contrast with Kirk (2008), subjects were instructed to try to recognize the objects and scenes were displayed before the target objects appeared. This mode of presentation is known to affect brain activity (Mudrik et al., 2010, 2014), likely because the scene has already been processed when the target object appears. It was hypothesized that PHC and RSC activity would correspond with congruent but not neutral contexts. Given the possibility of finding context-related activities in other regions of the brain, including the frontal and parietal areas, exploratory analyses were also conducted.

The influence of scenes on the recognition of objects was examined in this study by making objects appearing in scenes that were either congruent with the object or neutral. PHC and RSC activity emerged when objects were viewed in congruent scenes; however, activity was also observed in the right parietal lobe, fusiform gyrus and cerebellar regions. These findings help specify the role of PHC and RSC regarding the object-context relationship, and indicate that different brain regions are employed when context helps to identify an object.

Our findings compliment the current state of knowledge, the PHC activity is particular to place-information of object during recognition while episodic memory is employed via the RSC when viewing congruent contexts to give the context meaning. These two brain areas are more activated by scenes than by objects (Epstein and Kanwisher, 1998; Epstein and Higgins, 2006) preferentially respond to scenes relative to faces or single objects. As the present study demonstrates, spatial information processed through the PHC is crucial for object identification: objects may be difficult to recognize partially because of the angle they are viewed from and having information about the position of an object in space may provide enough information to elicit its recognition. For example, an object in front of a desktop computer will likely be a keyboard. Our findings are further substantiated by the spatial layout hypothesis which states that the PHC processes information about the shape of space embodied in layout-defining scene features (Epstein and Kanwisher, 1998).

In the well-known Bar’s model of object recognition, the PHC and RSC mediate the familiar associations of information at two levels of abstraction (Bar, 2004; Bar et al., 2008). The PHC was found to be activated by spatial contextual information, including places, associated to specific objects (Bar and Aminoff, 2003). Spatial contextual information is processed quickly and provides representations that facilitate and fasten object recognition they are associated with by allowing predictions about what could appear in the scene (Fenske et al., 2006; Gronau et al., 2008). According to Bar (2009), these predictions are made before object is recognized, even when the scene is perceived simultaneously, because they are computed from low-level information that are quickly conveyed and processed. In the present study, the associative representations activated by the scenes were already accessible when object was perceived. This may have bolstered the effect of scene, although it has been shown in previous studies on event-related potentials that presenting the scene before or simultaneously has not an important effect on scene influence of object processing (Ganis and Kutas, 2003).

The present PHC activities is consistent with other evidences indicating that this brain structure plays a critical in the integration of objects within their scenes. For instance, in episodic memory tasks, the PHC was shown to be more activated for objects that were perceived in scenes where they had been initially studied (Hayes et al., 2007). However, other studies may have contradictory results such as in Rémy et al. (2014), where the PHC was more activated for incongruent than congruent scene-objects. This effect was, however, more likely related to the processing of incongruity than to the beneficial influence of scene on the recognition of the object.

The associations mediated within the RSC are more general and amodal than those mediated in the PHC. Objects that are related to spatial or nonspatial contexts activate more strongly the RSC than objects with no specific context (Bar and Aminoff, 2003). Based on other studies, the RSC activity may be unrelated to the processing of objects and be alternatively involved in a larger network mediating the spatial aspects of scenes (Epstein et al., 2007; Harel et al., 2013). While the present studies cannot specify the exact role played by the RSC, it nevertheless shows that it participates to the influence that scenes exert on the recognition of objects.

The exploratory analyses showed activity in the IPL and fusiform gyrus, both of which may also contribute to context meaning. The IPL has been associated with autobiographical memory and its activity when viewing congruent contexts suggests that past experiences with similar contexts are being accessed (Berryhill et al., 2007). The IPL has also been shown to play a crucial role in spatial processing and recognition of salient information, and it may help apply past memories to interpret presented stimuli (Battelli et al., 2007; Husain and Nachev, 2007; Singh-Curry and Husain, 2009). Likewise, the fusiform gyrus may also facilitate application of contextual memories to object recognition. Although it is best known for facial processing, previous studies suggest that it also facilitates top-down processing of contextual information and differently engages when viewing visible compared with obscured objects (Haynes et al., 2005; Fenske et al., 2006; Gutchess et al., 2006).

In summary, three key findings emerged: PHC activity appears to be specific to object identification, while activity in the RSC, IPL and fusiform gyrus may facilitate the recall of contextual meanings when viewing congruent scenes.

MBB is the main supervisor. He funded the project, designed the experiment, supervised the data collection and data analyses and significantly contributed to the writing. DC has collected data, analyzed the data and wrote the article. CH has significantly helped with the study design and data analyses.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.