Learning to see faces and objects

Visual recognition of objects is an impressively difficult problem that biological systems solve effortlessly. We consider two aspects of this ability. First, is the recognition of all objects accomplished by either a single system or multiple, domain-specific systems? Behavioral, neuropsychological and neuroimaging data indicate that a single system is sufficient for the recognition of all objects at all levels. Second, how does such a system 'tune' itself to the constraints imposed by recognition at different levels of specificity? Evidence indicates that the task demands and learning that arise from different forms of feedback determine which computational routines are recruited automatically in object recognition.     

Familiarity effects on categorization levels of faces and objects

It is well established that faces, in contrast to objects, are categorized as fast or faster at the individual level (e.g., Bill Clinton) than at the basic-level (e.g., human face). This subordinate-shift from basic-level categorization has been considered an outcome of visual expertise with processing faces. However, in the present study we found that, similar to familiar faces, categorization of individually-known familiar towers is also faster at the individual level than at the basic-level in naïve participants. In addition, category-verification of familiar stimuli, at basic and superordinate levels, was slower and less accurate compared to unfamiliar stimuli. Thus, the existence of detailed semantic information, regardless of expertise, can induce a shift in the default level of object categorization from basic to individual level. Moreover, the individually-specific knowledge is not only more easily-retrieved from memory but it might also interfere with accessing more general category information.     

Competitor effects in naming objects and famous faces

The prior production of an alternative name increases the time taken to name a famous face. For example, naming a picture of the comedy actor “John Cleese” by the name of the character he played in the TV series Fawlty Towers (Basil Fawlty) increases the time required to subsequently produce the name “John Cleese”. This effect has been termed the “nominal competitor effect”. In contrast prior production of a property associated with a famous person has no effect on naming speed. For example, prior production of the name of the TV series Fawlty Towers does not slow subsequent production of “John Cleese”. The experiments reported explored analogous effects in object naming. Experiment 1 examined the effects of prior production of an alternative name (e.g., from American English or British English) and a semantic associate on the time taken to name line drawings of objects. It was found that prior production of an alternative name slowed object naming, but prior production of the name of a semantic associate did not. Experiment 2 demonstrated that cueing a specific name (e.g., the British English name) was not a necessary condition for the nominal competitor effect on object naming. Experiment 3 demonstrated that the nominal competitor effect on naming famous faces was also observed under both cued and uncued naming instructions. The data from both object and face naming are interpreted within the terms of current models of speech production.     

The fidelity of visual memory for faces and non-face objects

The fidelity of visual working memory was assessed for faces and non-face objects. In two experiments, four levels of memory load (1, 2, 3, or 4 items) were combined with four perceptual distances between probe and study items, with maximum item confusability occurring for the minimum memory load. Under these conditions, recognition memory for multiple faces exceeded that of a single face. This result was primarily due to the higher false alarm rates for faces than non-face objects, even though the two classes of stimuli had been matched for perceptual discriminability. Control experiments revealed that this counterintuitive result emerged only for old–new recognition choices based on near-threshold image differences. For non-face objects, instead, recognition performance decreased with increasing memory load. It is speculated that the low memorial discriminability of the transient properties of a face may serve the purpose of enhancing recognition at the individual-exemplar level.     

Search asymmetries with real faces: testing the anger-superiority effect

The anger-superiority hypothesis states that angry faces are detected more efficiently than friendly faces. Previously research used schematized stimuli, which minimizes perceptual confounds, but violates ecological validity. The authors argue that a confounding of appearance and meaning is unavoidable and even unproblematic if real faces are presented. Four experiments tested carefully controlled photos in a search-asymmetry design. Experiments 1 and 2 revealed more efficient detection of an angry face among happy faces than vice versa. Experiment 3 indicated that the advantage was due to the mouth, but not to the eyes, and Experiment 4, using upright and inverted thatcherized faces, suggests a perceptual basis. The results are in line with a sensory-bias hypothesis that facial expressions evolved to exploit extant capabilities of the visual system.     

Recognition of faces and complex objects in younger and older adults

We examined whether (1) age-associated impairments in face recognition are specific to faces or also apply to within-category recognition of other objects and (2) age-related face recognition deficits are related to impairments in encoding second-order relations and holistic information. In Experiments 1 and 2, we found reliable age differences for recognition of faces, but not of objects. Moreover, older adults (OAs) and younger adults (YAs) displayed similar face inversion effects. In Experiment 3, unlike YAs, OAs did not show the expected decline in performance for recognition of composites (Young, Hellawell, and Hay, 1987). In Experiment 4, both OAs and YAs showed a whole/part advantage (Tanaka and Farah, 1993). Our results suggest that OAs have spared function for processing of second-order relations and holistic information. Possible explanations for the finding that OAs have greater difficulty recognizing faces than recognizing other objects are proposed.     

Faces as objects of non-expertise: processing of thatcherised faces in congenital prosopagnosia

Congenital prosopagnosia (cPA) is a severe disorder in recognising familiar faces, a human characteristic that is presumably innate, without any macro-spatial brain anomalies. Following the idea that cPA is based on deficits of configural face processing, we used a speeded grotesqueness decision task with thatcherised faces, since the Thatcher illusion can serve as a test of configural disruption (Lewis and Johnston, 1997 Perception 26 225-227). The time needed to report the grotesqueness of a face in relation to orientation showed dissociate patterns between a group of fourteen people with cPA and a group of matched controls: whereas the RTs of controls followed a strong sigmoid function depending on rotation from the upright orientation, the RTs of people with cPA showed a much weaker sigmoid trend approaching a linear function. The latter result is interpreted as a diagnostic sign of impaired configural processing, being the primary cause of the absence of 'face expertise' in prosopagnosia.     

Semantic priming in the name retrieval of objects and famous faces

Researchers interested in face processing have recently debated whether access to the name of a known person occurs in parallel with retrieval of semantic‐biographical codes, rather than in a sequential fashion. Recently, Schweinberger, Burton, and Kelly (2001) took a failure to obtain a semantic context effect in a manual syllable judgment task on names of famous faces as support for this position. In two experiments, we compared the effects of visually presented categorically related prime words with either objects (e.g. prime: animal; target: dog) or faces of celebrities (e.g. prime: actor; target: Bruce Willis) as targets. Targets were either manually categorized with regard to the number of syllables (as in Schweinberger et al.), or they were overtly named. For neither objects nor faces was semantic priming obtained in syllable decisions; crucially, however, priming was obtained when objects and faces were overtly named. These results suggest that both face and object naming are susceptible to semantic context effects.     

Seeing faces as objects: no face inversion effect with geometrical discrimination

Inversion dramatically impairs face perception, recognition, and discrimination. Yet it does not interfere with the ability to make precise estimates of facial feature distances. To investigate this discontinuity between facial feature distance estimation and general perception and recognition, we assessed the effect of inversion on the discrimination of differences in facial compression and elongation or expansion using geometrically distorted faces. The results clearly showed that geometrical face discrimination is not subject to the traditional face inversion effect and did not show a benefit for natural faces. Although discrimination thresholds were not affected by inversion, response times to the distance judgments were faster with inversion, especially when the inverted faces contained natural configurations. Based on these counterintuitive results, we suggest that participants used analytical processing to do the discrimination task. Moreover, we suggest that the depth with which a face is holistically encoded depends on the nature of the task, face orientation, and similarity between a face and the prototypical face template.     

Seeing big things: overestimation of heights is greater for real objects than for objects in pictures

In six experiments we demonstrate that the vertical-horizontal illusion that is evoked when viewing photographs and line drawings is relatively small, whereas the magnitude of this illusion when large objects are viewed is at least twice as great. Furthermore, we show that the illusion is due more to vertical overestimation than horizontal underestimation. The lack of a difference in vertical overestimation between pictures and line drawings suggests that vertical overestimation in pictures depends solely on the perceived physical size of the projection on the picture surface, rather than on what is apparent about an object's represented size. The vertical-horizontal illusion is influenced by perceived physical size. It is greater when viewing large objects than small pictures of these same objects, even when visual angles are equated.     

Irrelevant objects of expertise compete with faces during visual search

Prior work suggests that nonface objects of expertise can interfere with the perception of faces when the two categories are alternately presented, suggesting competition for shared perceptual resources. Here, we ask whether task-irrelevant distractors from a category of expertise compete when faces are presented in a standard visual search task. Participants searched for a target (face or sofa) in an array containing both relevant and irrelevant distractors. The number of distractors from the target category (face or sofa) remained constant, whereas the number of distractors from the irrelevant category (cars) varied. Search slopes, calculated as a function of the number of irrelevant cars, were correlated with car expertise. The effect was not due to car distractors grabbing attention, because they did not compete with sofa targets. Objects of expertise interfere with face perception even when they are task irrelevant, visually distinct, and separated in space from faces.     

Memory for objects in real scenes: The development of recognition and recall

Subjects examined crowded semirealistic layouts of toy objects, or photographs of these layouts, and then tried to identify added, moved or deleted items. The main study, involving 1st-, 3rd, 6th-graders, and adults, showed that although successful recognition of added items was superior to both recall of deleted items and recognition of moved ones, all three scores improved with age. In addition, false reports of “new” items decreased markedly in the older groups. The results argue against the widely held view that recognition memory undergoes little or no developmental improvement. No significant difference between real layouts and photographs appeared either in the main experiment or in replications involving shorter exposure (Experiment 2) or retarded subjects (Experiment 3).     

Attentional and perceptual factors affecting the attentional blink for faces and objects

When 2 different visual targets presented among different distracters in a rapid serial visual presentation (RSVP) are separated by 400 ms or less, detection and identification of the 2nd targets are reduced relative to longer time intervals. This phenomenon, termed the attentional blink (AB), is attributed to the temporary engagement of a limited-capacity attentional system by the 1st target, which reduces resources available for processing the 2nd target. Although AB has been reliably obtained with many stimulus types, it has not been found for faces (E. Awh et al., 2004). In the present study, the authors investigate the underpinnings of this immunity. Unveiling circumstances in which AB occurs within and across faces and other categories, the authors demonstrate that a multichannel model cannot account for the absence of AB effects on faces. The authors suggest instead that perceptual salience of the face within the distracters' series as well as the available resources determine whether or not faces are blinked in RSVP.     

Perceptual asymmetries in left- and right-handers for cartoon and real faces

We examined 40 left-handers and 40 right-handers on two free-vision tests of face processing. A chimeric face composed of a smiling half-face joined to either a neutral half-face (real faces) or a sad half-face (cartoon faces) and its mirror image were presented on each trial. Subjects judged which chimeric face looked happier, the one with the smile to the left or the one with the smile to the right. Right-handers, but not left-handers, had a highly significant leftward attentional bias, since chimeras with the smile to the left were judged happier than those with the smile to the right. The cartoon- and real-face tasks did not differ in the mean perceptual asymmetries they elicited. However, correlations between attentional-asymmetry scores for the two tasks, although high for both left- and right-handers, were significantly smaller than task reliabilities. Thus, the cartoon- and real-face tasks overlap to a major extent in the lateralized processes they measure, but to a lesser extent, they also index different lateralized processes.     

Effect of scene complexity on colour constancy with real three-dimensional scenes and objects

The effect of scene complexity on colour constancy was tested with a novel technique in which a virtual image of a real 3-D test object was projected into a real 3-D scene. Observers made discriminations between illuminant and material changes in simple and complex scenes. The extent of colour constancy achieved varied little with either scene structure or test-object colour, suggesting a dominant role of local cues in determining surface-colour judgments.     

First impressions from faces in the real world: Commentary on Sutherland and Young (2022)

The study of first impressions from faces now emphasizes the need to understand trait inferences made to naturalistic face images (British Journal of Psychology, 113, 2022, 1056). Face recognition algorithms based on deep convolutional neural networks simultaneously represent invariant, changeable and environmental variables in face images. Therefore, we suggest them as a comprehensive ‘face space’ model of first impressions of naturalistic faces. We also suggest that to understand trait inferences in the real world, a logical next step is to consider trait inferences made to whole people (faces and bodies). On the role of cultural contributions to trait perception, we think it is important for the field to begin to consider the way in which trait inferences motivate (or not) behaviour in independent and interdependent cultures.     

Talking emotions: vowel selection in fictional names depends on the emotional valence of the to-be-named faces and objects

One prestudy based on a corpus analysis and four experiments in which participants had to invent novel names for persons or objects (N?=?336 participants in total) investigated how the valence of a face or an object affects the phonological characteristics of the respective novel name. Based on the articulatory feedback hypothesis, we predicted that /i:/ is included more frequently in fictional names for faces or objects with a positive valence than for those with a negative valence. For /o:/, the pattern should reverse. An analysis of the Berlin Affective Word List – Reloaded (BAWL-R) yielded a higher number of occurrences of /o:/ in German words with negative valence than in words with positive valence; with /i:/ the situation is less clear. In Experiments 1 and 2, participants named persons showing a positive or a negative facial expression. Names for smiling persons included more /i:/s and fewer /o:/s than names for persons with a negative facial expression. In Experiments 3 and 4, participants heard a Swahili narration and invented pseudo-Swahili names for objects with positive, neutral, or negative valence. Names for positive objects included more /i:/s than names for neutral or negative objects, and names for negative objects included more /o:/s than names for neutral or positive objects. These finding indicate a stable vowel-emotion link.     

Dissociating positive and negative influences of verbal processing on the recognition of pictures of faces and objects

Four experiments investigated the role of verbal processing in the recognition of pictures of faces and objects. We used (a) a stimulus-encoding task where participants learned sequentially presented pictures in control, articulatory suppression, and describe conditions and then engaged in an old-new picture recognition test and (b) a poststimulus-encoding task where participants learned the stimuli without any secondary task and then either described or not a single item from memory before the recognition test. The main findings were as follows: First, verbalization influenced picture recognition. Second, there were contrasting influences of verbalization on the recognition of faces, compared with objects, that were driven by (a) the stage of processing during which verbalization took place (as assessed by the stimulus-encoding and poststimulus-encoding tasks), (b) whether verbalization was subvocal (whereby one goes through the motions of speaking but without making any sound) or overt, and (c) stimulus familiarity. During stimulus encoding there was a double dissociation whereby subvocal verbalization interfered with the recognition of faces but not objects, while overt verbalization benefited the recognition of objects but not faces. In addition, stimulus familiarity provided an independent and beneficial influence on performance. Post stimulus encoding, overt verbalization interfered with the recognition of both faces and objects, and this interference was apparent for unfamiliar but not familiar stimuli. Together these findings extend work on verbalization to picture recognition and place important parameters on stimulus and task constraints that contribute to contrasting beneficial and detrimental effects of verbalization on recognition memory.