FACE PROCESSING - FACE RECOGNITION PROBLEMS

Vocabulary flashcards covering prosopagnosia, its subtypes, core theories of face recognition, neural substrates, covert recognition evidence, and related tests and phenomena discussed in the lecture notes.

Acquired prosopagnosia

Prosopagnosia resulting from brain damage (often bilateral, sometimes right-dominant) affecting face recognition despite intact object recognition and other skills.

Developmental (Congenital) prosopagnosia

Prosopagnosia present from birth or early childhood without brain damage; may run in families and involve deficits in configural/holistic face processing.

Cognitive Evidence for Specialist Face Processing

Phenomena that highlight the holistic nature of face processing

• Face Inversion Effect,

• Thatcher Illusion

• Composite Face Effect,

which suggest a dedicated system for recognizing upright faces.

Face Inversion Effect

• Faces are harder to recognize when inverted compared to upright.

• Upright faces are processed holistically, inverted faces are processed by parts.

• Evidence: Normal participants recognize upright faces with ~94% accuracy but only ~82% when inverted.

• Prosopagnosia patient (left hemisphere lesion) shows impaired recognition for upright (58%) but slightly better for inverted (72%), consistent with disrupted holistic processing.

Thatcher Illusion

• When a face’s eyes and mouth are inverted but the whole face is upside down, it looks normal.

• When the face is upright, the inverted features look grotesque or strange.

• Reveals that face processing relies heavily on holistic perception in the upright orientation.

Composite Face Effect

• People use more holistic processing for upright faces than inverted ones.

• When two halves of different faces are aligned upright, they are harder to recognize as separate individuals.

• When inverted, it’s easier to recognize the two halves belong to different people.

Patient Faran’s case and word and object recognition?

• Patient had acquired alexia (letter-by-letter reading, confusing similar words) from a left hemisphere lesion.

• This is a type of visual agnosia affecting word recognition, often alongside object recognition problems.

model predictiond of face, object, and word recognition deficits

• Face recognition relies on holistic processing (overall configuration), object and word recognition rely on decomposing parts.

• Patients can have word recognition problems without face/object problems, or face recognition problems without word/object problems.

• Co-occurrence is expected for faces & objects and objects & words, but not faces & words together because they use different processes.

why impaired face recognition but intact object recognition?

• It could be due to different sensitivity levels of the tests used for face vs. object recognition.

• Object recognition tests often involve between-category distinctions (e.g., clothes vs. tools).

• Face recognition involves distinguishing within the same category (all faces), which is more challenging.

evidence separating prosopagnosia from other agnosias

• Because faces are all from the same category, impairments in face recognition may reflect within-class recognition difficulty.

• Object recognition usually involves between-class distinctions, making it easier to detect deficits.

• This raises questions about how clearly prosopagnosia is dissociable from other forms of agnosia.

Holistic processing

Processing that integrates facial features into a single overall representation rather than as independent parts.

Configural processing

Processing of the spatial arrangement of facial features; essential for holistic face recognition.

sheep farmer’s case and face and object recognition?

Suggests face recognition can be impaired independently from object recognition.

Patient LH memory test

• Study phase: Half the stimuli (faces and objects like glasses) presented.

• Test phase: All stimuli presented (half studied, half new).

• Task: Judge if stimulus is old or new.

Patient LH’s memory test performance compared to controls

• Patient LH: 64% correct for faces, 63% for glasses (objects).

• Controls: 85% correct for faces, 69% for glasses.

• Shows impaired face recognition but relatively preserved object recognition, supporting dissociation.

cognitive evidence supporting a specialist face processing system

• People recognize faces better when processing features holistically (as a whole).

• People recognize objects better when processing features separately (part by part).

brain evidence for a special system for recognizing faces?

• Problems with recognizing faces usually happen when both sides of the brain are damaged (from things like strokes or head injuries).

• Studies in monkeys found brain cells that only respond to faces.

• Sometimes face and object problems happen together, but face recognition needs special brain areas. (fusiform face area in humans)

single cell recordings in monkeys and face processing

• Certain brain cells respond strongly to monkey faces and human faces, especially frontal monkey profiles.

• Some cells respond to all kinds of facial features.

• However, we can’t say these cells only process faces for sure.

Early Face Recognition in Infants

• Newborns can tell their mother’s face from others by 2 days old.

• Babies a few months old recognize familiar faces.

• Rapid face recognition development suggests it might be a special brain module.

Bruce and Young model

Cognitive model of face recognition with stages such as Structural Encoding, Viewer-Centered Descriptions, Expression-Independent Descriptions, Face Recognition Units, Person Identity Nodes, semantic information, and naming.

Structural Encoding Stage

Early stage that extracts basic facial features; damage can cause apperceptive prosopagnosia.

• includes viewer centered descriptions and expression independent descriptions

Viewer-Centered Descriptions

• They are viewpoint-specific representations of a face.

• Help recognize facial expressions, understand speech, and judge differences like unfamiliar people, gender, and age.

• Focus on raw perceptual features like surface shapes, lumps, and bumps.

TESTS

Differentiate between faces and objects: Point to the picture of a face

Pick out features: Point to the eyes

Expression Independent Descriptions

• They encode the face’s global layout and specific features from different viewpoints.

• Allow you to mentally put the face together as a whole (holistic processing).

• Help you mentally rotate the face and recognize it regardless of the angle or expression.

• Focus on the overall configuration, not just individual features.

TESTS

Match faces from different perspectives: Are these two faces the same or different?

Face perception test: Sort the faces on the bottom line in order of similarity to the top face (work from most to least similar)

Differentiate faces and scrambled faces: Point to the face shown below

Face Recognition Units (FRUs)

• FRUs receive input from Expression Independent Descriptions.

• They store visual structural descriptions of familiar faces.

• Activate when a viewed face strongly matches a stored face representation.

• Separate FRU exists for each known face.

• Can also be triggered by semantic info from Person Identity Nodes.

• Respond to known faces from any angle but not to voice or other context clues.

• If FRUs are damaged, a person cannot recognize familiar faces properly (e.g., famous people).

TESTS

Recognise Familiar Faces: Is this person familiar?

Prime familiarity decision: Look at the picture and then tell me the name you see

Face memory test: Remember the faces below, Which face did you just view

Person Identity Nodes (PINs)

• PINs store semantic (biographical) information about a person, such as occupation and characteristics.

• They activate after Face Recognition Units (FRUs) recognize the face.

• If PINs are damaged, a person can recognize someone’s face but cannot recall any personal information about them (e.g., occupation, characteristics)

TESTS

Politician test: Is the name that of a Politician

Bibliographic information: Tell me the occupation of the person in the picture?

Name Generation Stage

• After the Person Identity Nodes activate, semantic info allows you to retrieve the person’s name.

• To test for naming deficits specific to faces, patients are shown other pictures (like cities or countries).

• If they fail to name faces but can name non-face items, it indicates a naming deficit specific to people’s faces.

TESTS

Name faces: Tell me the name of the person in the photograph

Name cities/countries: Tell me the name of the country associated with this picture

Expression Analysis

• It involves identifying facial expressions separately from recognizing the face itself.

• Helps in recognizing emotions by analyzing facial movements and features.

TESTS

Interpret facial expression

Facial Speech Analysis

• It involves using lip and tongue movements to help understand speech (like lip reading).

• If this system is impaired, people won’t experience visual-auditory interference—they’ll rely only on what they hear, not visual cues.

TESTS

McGurk effect

Lip reading

Directed Visual Processing

• It involves focusing attention on specific face features to learn new faces and compare faces feature-by-feature.

• A deficit means difficulty learning new faces because the person can’t analyze or encode facial features properly.

Cognitive System

• It involves higher-order cognitive processes like episodic memory, attention, and decision-making.

• Supports recognizing faces by integrating memories and helping make recognition decisions.

Patient PG’s case and prosopagnosia

• Damage to the right side of the brain caused problems with perceptual processing of faces.

• couldn’t tell a normal face from a scrambled one, showing difficulty in structural encoding.

• couldn’t form viewer-centered or expression-independent face descriptions, impairing recognition.

Patient PH’s and prosopagnosia

• Occipitotemporal lesions caused difficulty in recognizing faces.

• could tell faces from non-faces but couldn’t judge face familiarity.

• couldn’t access semantic information about faces.

• Shows a problem at the Face Recognition Units stage: perceiving but not recognizing faces.

Patient ME’s case and prosopagnosia

• could distinguish faces from non-faces and rate face familiarity accurately.

• could match photos of the same face from different angles.

• However, could NOT recall any semantic information about the recognized faces — just a feeling of familiarity.

• Indicates a problem after Face Recognition Units but before semantic access (likely at the Person Identity Nodes stage).

Patient EST’s and prosopagnosia

• Partial removal of left temporal lobe caused anomia (difficulty naming faces).

• could process faces, rate familiarity, and recall semantic information.

• However, could NOT name the faces, showing a problem at the Name Generation stage.

Burton, Bruce and Johnston Face Recognition Model

• It’s an interactive network with three main pools:

  1. Face Recognition Units (FRUs)

  2. Person Identity Nodes (PINs)

  3. Semantic Information Units

Burton, Bruce and Johnston Face Recognition Model Interaction

• Within each pool, units inhibit each other (competition), while pools excite each other (collaboration).

• FRUs activate when any view of a familiar face is seen and excite PINs.

• PINs link to semantic units containing knowledge about the person (e.g., movies an actor was in).

Overt recognition

conscious, deliberate face recognition.

Covert recognition

unconscious or automatic face

can face processing happen without awareness

• processing, even without conscious awareness.

• Prosopagnosia patients may lack overt recognition but still show covert face processing.

• Physiological and behavioral studies support face processing can occur without awareness.

Patient LF and covert face recognition

• Could not consciously identify familiar faces (performed at chance).

• Showed greater and more frequent Skin Conductance Responses (SCRs) to correct names matched with familiar faces, indicating unconscious recognition.

Tranel and Damasio Study- physiological evidence for covert face recognition?

Patient’s SCRs increased when shown familiar faces compared to unfamiliar faces, showing unconscious emotional response.

Rizzo et al. Study- eye movement tracking and prosopagnosic

Prosopagnosics looked at similar facial features on familiar faces as healthy controls, suggesting unconscious processing of faces.

Renault et al. Study- covert face recognition

Found increased P300 amplitude (event-related potentials) for familiar faces during a visual oddball task, indicating brain response to familiarity even without conscious recognition.

Patient PH – Behavioral Evidence of covert face recognition

• could not consciously recognize familiar faces and performed at chance on forced-choice tests.

• Despite this, showed covert recognition in:

  • Face matching tasks

  • Interference paradigms

  • Associative priming tasks

Patient PH’s and Face Matching Task

• Two faces are shown simultaneously, often in different orientations.

• The task is to decide if the faces are of the same person, requiring holistic face processing.

• Control participants are faster at matching familiar faces than unfamiliar ones.

• Patient also shows this familiarity effect, indicating covert face recognition despite not consciously recognizing faces.

Patient PH’s and Interference Task

• Faces are shown with a printed name that can be:

  • The actual person’s name

  • Name of someone with a related occupation

  • An unrelated person’s name

• Participants decide if the name matches the face.

• Controls take longer to respond when the name is unrelated, showing interference.

• Patient PH also shows this interference effect, indicating covert recognition despite no conscious face recognition.

Patient PH’s and Associative Priming Task

• A prime face is shown before a written target name.

• Task: decide if the target name is familiar.

• Controls respond faster when the prime face is associated with the target name, compared to unrelated or unfamiliar primes.

• showed the same effect, indicating covert face recognition despite conscious recognition deficits.

Superior Temporal Sulcus (STS)

Brain area with cells responsive to faces and dynamic social cues; implicated in face processing in primates.

Anomia for people's names

Selective naming deficit where a person cannot name familiar people but can describe them or recognize them.

• Patient GBL could only name 3 out of 20 famous people shown in photos.

• Despite this, GBL could accurately describe who 18 out of 20 people were.

• Patient could correctly name European and English towns, showing no general noun naming deficit.

• Indicates a selective naming deficit specifically for people’s names.

Face anosognosia

Lack of awareness of one’s own face recognition deficit, sometimes after right-hemisphere damage.

• Could name only 3/20 famous people’s photos.

• Could accurately describe 18/20 people’s identities.

• Could name European and English towns correctly.

• Shows a selective naming deficit specific to people’s names, not a general noun deficit.

Metamorphosia- Perceptual Distortion

Distorted perception of faces, with distorted proportions (e.g., nose or mouth) reported by patients with face processing impairments.

Patient JS and processing issues after a stroke

• Problems recognizing family members, including her own daughter she sees regularly.

• No trouble recognizing daughter she hadn’t seen for 8 years.

• Found family faces repulsive; distorted facial proportions perceived.

• Impaired facial expression perception and poor performance on Benton Facial Recognition Test.

• Slower and less accurate at identifying celebrity and family faces compared to controls.

• Skin conductance responses (SCR) showed mixed results but indicated lower arousal overall.

• Not prosopagnosia or Capgras delusion; possibly mild metamorphosia affecting identity and emotion integration.

Super-recognisers

Individuals with exceptional lifelong face recognition abilities, performing at or above controls on memory and perception tests; evidence for a continuum of face recognition skills.

• super recognisers were more accurate in the Cambridge Face Memory test than the control participants

Cambridge Face Memory Test (CFMT)

Standardized test of face memory across multiple viewpoints used to identify strong face recognition ability.

• Learn to recognize 6 unfamiliar faces from 3 different perspectives.

• Test phase: identify the trained faces from a set of images showing one of the three learned views.

• Task difficulty increases as images change in orientation, lighting, and quality.

Before They Were Famous (BFWF) test

Test assessing recognition of famous faces from various perspectives, used to gauge facial memory.

Cambridge Face Perception Test key findings

• Task: order faces from most to least similar to a target face.

• Faces are shown upright or inverted to increase difficulty.

• Super recognisers perform better than controls and prosopagnosics with upright faces.

• When faces are inverted, super recognisers lose their advantage because holistic face processing is disrupted.