PSY1004: Cognitive Psychology - Lecture 3: Perception

Foundations of Sensation and Perception

• Definition of Perception: The cognitive experience resulting from stimulation of the senses. It comprises a complex set of operations by which sensory input from environmental stimuli is organized, recognized, and interpreted into meaningful experiences.

• Dynamic and Adaptive Nature: Perception is not static; it constantly updates and modifies based on newly added information from the environment.

• Active Cognitive Operation: Rather than passive reception, perception is an active process analogous to reasoning, hypothesis testing, and problem solving.

• Perception-Action Coupling: Perceptual processing operates in direct conjunction with physical actions, continuously guiding movement while movement updates incoming sensory signals.

• Dual Information Sources in Perception:

  • Environmental energy: Physical stimulation of sensory receptors by external stimuli (data-driven input).

  • Observer knowledge: Stored knowledge, past experiences, and expectations that the perceiver brings to the situation (concept-driven input).

• The Scope of Human Senses:

  • Classical five senses: Sight (vision), hearing (audition), smell (olfaction), touch (somatosensation), and taste (gustation).

  • Contemporary sensory physiology: Humans possess an estimated $22$ to over $30$ distinct sensory modalities.

  • Thermoreception: The detection of temperature changes (heat and cold).

  • Vestibular sense: Sensation of balance, gravitational pull, and head orientation in space.

  • Proprioception: The awareness of body posture, limb position, and muscle tension without visual input.

  • Nociception: Sensory detection of noxious, potentially tissue-damaging stimuli experienced as pain.

  • Interoception: Perception of internal physiological states and visceral signals, including hunger, thirst, heart rate, respiration, and fatigue.

  • Vibration and pressure reception: Specialized mechanoreception tuned to rapid tactile oscillations and sustained mechanical indentation.

• Cortical Hubs of Primary Sensory Processing:

  • Visual cortex: Located in the occipital lobe, dedicated to visual feature processing.

  • Auditory cortex: Situated in the superior temporal lobe, dedicated to acoustic processing.

  • Olfactory cortex: Processes airborne chemical odorants; connects to the limbic system and bypasses the obligate thalamic sensory relay.

  • Gustatory cortex: Located within the insular cortex and frontal operculum, processes tastants.

  • Somatosensory cortex: Positioned along the postcentral gyrus of the parietal lobe, maps somatic touch, temperature, and nociceptive signals.

  • Thalamus: Acts as the major subcortical sensory relay center routing nearly all ascending sensory data to designated cortical destinations.

  • Brain stem: Manages ascending equilibrium signals, autonomic integration, and early acoustic processing.

Cortical Topography: The Sensory and Motor Homunculi

• The Cortical Homunculus: An anatomical map representing the somatotopic organization of the human body across both the primary somatosensory cortex (postcentral gyrus) and the primary motor cortex (precentral gyrus).

• Medial-to-Lateral Topographical Mapping:

  • Medial longitudinal fissure: Genitalia / genitals, toes, foot, and leg.

  • Superior crest: Hip, trunk, neck, and head.

  • Upper extremity trajectory: Shoulder, arm, elbow, forearm, wrist, hand, and fingers (ordered little, ring, middle, index, and thumb).

  • Lateral craniofacial trajectory: Brow, eye, nose, face, upper lip, lips, lower lip, teeth, gums, jaw, tongue, pharynx, larynx (motor control), and intra-abdominal organs / abdomen (sensory mapping).

• Principle of Cortical Magnification: The amount of cortical surface area allocated to a body part is not proportional to its physical size, but corresponds directly to its receptor density, sensory sensitivity, and the degree of motor control required. Areas such as the hands, fingers, lips, and tongue occupy massively disproportionate expanses of cortical tissue relative to the trunk, hips, and legs.

The Constructive Nature of Perceptual Objects

• Distal Stimulus vs. Perceptual Object:

  • Distal object: The physical entity existing out in the objective environment (for example, a cat sitting in ambient light).

  • Proximal stimulus: The physical energy pattern impinging directly upon sensory receptors (such as light rays striking the retina).

  • Perceptual object: The internal neurocognitive representation and reconstruction synthesized by the brain. It is an interpretation and reproduction of the distal object, rather than a direct, photographic reflection.

• Neural Correlates of Consciousness (NCC): The specific minimal neural mechanisms, activations, and causal pathways within the brain that correspond directly to a specific subjective conscious percept.

Comparative Sensory Perception and Animal Vision

• Interspecies Perceptual Diversity: Human perception captures only a tiny slice of the physical world. Other biological species possess sensory systems tuned to physical properties that are completely invisible to humans.

• Ultraviolet and Infrared Detection:

  • Ultraviolet ($\text{UV}$) radiation is visible to most birds, insects, numerous fish species, and select mammals, but invisible to humans.

  • Infrared ($\text{IR}$) and thermal radiation are detected by pit vipers through facial pit organs; dogs and foxes also appear capable of detecting thermal heat signatures.

  • Goldfish possess a visual system capable of simultaneously detecting both ultraviolet and infrared wavelengths.

• Comparative Color Vision (Humans vs. Birds):

  • Human Trichromatic Vision: Relies on three cone photoreceptors sensitive to wavelengths across the visible spectrum:

    • Short-wavelength cone: peak at approximately $424\,nm$

    • Medium-wavelength cone: peak at approximately $530\,nm$

    • Long-wavelength cone: peak at approximately $560\,nm$

  • Bird Tetrachromatic Vision: Operates via four cone photoreceptor classes extending into the near-ultraviolet range:

    • Ultraviolet cone: peak at approximately $370\,nm$

    • Short-wavelength cone: peak at approximately $445\,nm$

    • Medium-wavelength cone: peak at approximately $508\,nm$

    • Long-wavelength cone: peak at approximately $565\,nm$

  • Ecological Functions of Avian UV Vision:

    • Plumage signaling: Plumage patterns that appear dull black or brown to human eyes display vivid, iridescent coloration under avian UV vision, aiding in courtship and mate selection.

    • Egg recognition: Ground-nesting bird eggs that blend into gravel under human vision stand out with distinctive contrast and pigmentation under UV vision.

• The Visual System of the Mantis Shrimp (Stomatopoda):

  • Possesses the most complex visual apparatus known in biology.

  • Contains $16$ distinct types of photoreceptors ($12$ dedicated to color channels spanning ultraviolet to deep red, and $4$ specialized for polarized light analysis).

  • Features two separate types of stereoscopic depth vision per eye, yielding independent trinocular vision in each eye.

• Hidden Human Morphology: Blaschko Lines:

  • Blaschko lines represent pathways of embryonic cell migration on human skin that follow invariant, zebra-like swirled striping patterns.

  • They are completely invisible under normal human trichromatic illumination because human skin exhibits uniform reflectance across the $400\,nm$ to $700\,nm$ range.

  • They are readily detectable to animals possessing ultraviolet vision, suggesting UV-sensitive species may identify human individuals based on distinctive stripe patterns.

  • Blaschko lines can become visible under high-intensity ultraviolet light or in clinical conditions involving genetic mosaicism or chimerism.

  • Morphological distributions across the body follow six distinct types: Type 1a, Type 1b, Type 2, Type 3, Type 4, and Type 5.

• The Electromagnetic Spectrum:

  • Spans frequencies from high-energy gamma rays ($10^{24}\,Hz$ to $10^{20}\,Hz$, wavelengths $10^{-16}\,m$ to $10^{-12}\,m$), X-rays ($10^{18}\,Hz$, $10^{-10}\,m$), and ultraviolet ($10^{16}\,Hz$, $10^{-8}\,m$), down to infrared ($10^{14}\,Hz$ to $10^{12}\,Hz$, $10^{-6}\,m$ to $10^{-4}\,m$), microwaves ($10^{10}\,Hz$, $10^{-2}\,m$), and broadcast radio waves ($10^{8}\,Hz$ to $10^{0}\,Hz$, $1\,m$ to $10^8\,m$).

  • The narrow visible window accessible to human eyes ranges only from violet at roughly $400\,nm$ to red at roughly $750\,nm$.

Challenges in Perception: Human Vision versus Artificial Systems

• Limitations of Machine Perception: Despite substantial advances, artificial computer vision systems encounter persistent errors when processing ambiguous natural scenes, highlighting fundamental computational challenges that human biology solves effortlessly.

• The Inverse Projection Problem:

  • The task of determining the exact three-dimensional distal stimulus responsible for a specific two-dimensional pattern of light projected onto the retina.

  • A two-dimensional retinal image can be mathematically generated by an infinite number of different physical objects of varying sizes, distances, shapes, and planar slants.

  • Because the problem is underdetermined, the visual system must use prior knowledge and environmental assumptions to select the correct 3D interpretation.

• Occlusion and Image Degradation:

  • Computer vision algorithms identify objects primarily by calculating edge boundaries through contrast differences between adjacent pixels.

  • When an object is partially occluded, blurry, or low in contrast, machine algorithms cannot reliably interpolate missing contours.

  • Human perceivers effortlessly fill in missing information using context, Gestalt grouping, and top-down schemas.

• Viewpoint Invariance:

  • The capacity of the visual cognitive system to identify an object reliably across radically different viewpoints, viewing angles, distances, and lighting conditions.

  • While humans maintain viewpoint invariance effortlessly, computational systems often misclassify objects when presented in non-canonical orientations.

• Adversarial Camouflage: CV Dazzle (Anti-Facial-Recognition Makeup):

  • Computer vision facial recognition pipelines depend on detecting spatial symmetry, the nasal bridge, cheekbone shadows, and eye-socket contours.

  • High-contrast, asymmetrical makeup patterns (CV Dazzle) disrupt pixel gradients and break facial symmetry, preventing facial recognition software from identifying and parsing the human face.

Processing Strategies: Bottom-Up versus Top-Down Processing

• Bottom-Up Processing (Direct Perception):

  • Data-driven processing originating directly from environmental stimuli stimulating the peripheral sensory receptors.

  • Sensory features (lines, colors, motion) are registered, assembled, and propagated upward through hierarchical neural stages to form a percept without requiring prior knowledge or contextual hypotheses.

• Top-Down Processing (Constructive Perception):

  • Concept-driven processing originating at high cognitive levels within the brain.

  • Perceptions are actively constructed using the observer's prior knowledge, contextual expectations, stored schemas, and hypotheses about the environment.

• Real-Time Integration: Human perception is neither exclusively bottom-up nor top-down; it involves the simultaneous, bidirectional interaction of both processing streams.

• Visual Search Paradigms:

  • Feature search (pop-out effect): When a target differs from distractors by a single salient visual primitive (such as color hue or orientation), it automatically "pops out." This is driven by preattentive, parallel bottom-up processing, with reaction times staying flat regardless of display set size.

  • Conjunction search: When a target is defined by a combination of two or more visual features (such as searching for a red circle among red squares and blue circles), the pop-out effect fails. The perceiver must deploy top-down attentional control to conduct a serial search across individual items or color-restricted clusters.

• Top-Down Modulation in Pain Perception:

  • Direct pathway model (bottom-up view): Posited that pain is a direct reflection of physical tissue damage, driven strictly by nociceptive stimulation traveling along an ascending pathway from the skin to the brain.

  • Cognitive modulation (top-down view): Pain intensity is heavily influenced by expectations, attentional focus, contextual framing, placebo effects, and emotional state. Distracting stimuli and positive expectations significantly attenuate subjective pain intensity despite constant nociceptive input.

• Top-Down Processing in Language and Speech:

  • The McGurk Effect: A cross-modal perceptual illusion where visual lip movements conflicting with an acoustic syllable cause the brain to perceive a completely different, fused phoneme.

  • Speech segmentation: The ability to perceive distinct word boundaries within a continuous auditory stream. The physical speech signal contains virtually no silent gaps between words; segmentation relies entirely on top-down linguistic knowledge.

  • Transitional probabilities: The statistical likelihood that one specific phoneme or syllable will follow another within a language, learned implicitly and used by the brain to anticipate upcoming sounds.

Theories of Object Perception: From Helmholtz to Gestalt Psychology

• Helmholtz’s Theory of Unconscious Inference:

  • Developed by Hermann von Helmholtz to resolve the ambiguity of the inverse projection problem.

  • States that our visual percepts are the product of rapid, automatic, unconscious assumptions that the cognitive system makes about environmental structures.

  • The Likelihood Principle: The visual system resolves ambiguous sensory input by generating the perceptual interpretation that is most likely to have produced that pattern of stimulation, based on past experience.

• Structuralism vs. Gestalt Psychology:

  • Structuralism (Wilhelm Wundt): Held that perception is built bottom-up by adding together basic elemental sensory units ("sensations").

  • Gestalt Psychology (Max Wertheimer, Wolfgang Köhler, Kurt Koffka): Rejected structuralism, arguing that the perceptual whole is qualitatively different from the sum of its parts ($\text{The whole is different from the sum of its parts}$).

  • Gestalt theorists maintained that the mind organizes visual input according to intrinsic, built-in laws of perceptual grouping rather than piecemeal learning.

• Gestalt Laws of Perceptual Organization:

  • Figure-ground segregation: The fundamental perceptual requirement to separate a primary visual object (the figure) from the background field (the ground). Exemplified by Rubin's ambiguous vase/faces illusion.

  • Principle of good continuation: Points that form straight or smoothly curving trajectories are perceived as belonging together; lines are seen as following the smoothest pathway.

  • Law of Prägnanz (Principle of Simplicity or Good Figure): The central Gestalt law asserting that every stimulus pattern is perceived in such a manner that the resulting structure is as simple, regular, and symmetrical as possible (for example, perceiving five overlapping Olympic rings rather than nine complex, irregular shapes).

  • Principle of similarity: Stimuli that share physical features (such as color, shape, size, or orientation) are automatically grouped together.

  • Principle of proximity: Visual elements located close to one another are perceived as belonging to a collective perceptual unit.

  • Principle of closure: The visual system fills in gaps to perceive an incomplete or open contour as a complete, unified figure.

  • Principle of common fate: Elements that move in the same direction and at the same speed are grouped together as a single entity.

  • Principle of symmetry: Visual regions bounded by symmetrical borders are preferentially perceived as cohesive figures.

• Gestalt Stance on Experience:

  • Gestalt psychologists viewed organizational principles as intrinsic, biologically built-in features of sensory processing (bottom-up).

  • Past experience was considered a secondary factor that could influence perception on occasion, but was not the primary driver of visual organization.

Environmental Regularities and Modern Predictive Perception

• The Modern Empirical Stance: Modern cognitive psychology recognizes that perceptual organization is not solely hardwired by innate laws, but is deeply shaped by adaptation to common physical and statistical regularities of the environment.

• Physical Regularities:

  • The Oblique Effect: Humans perceive horizontal and vertical lines with greater speed, accuracy, and acuity than diagonal (oblique) lines. This perceptual bias reflects the statistical predominance of vertical and horizontal structures in both natural and built environments, shaping cortical receptive fields via experience-dependent plasticity.

  • The Light-from-Above Assumption: The brain operates on the default heuristic that light comes from an overhead source (consistent with the sun and ceiling lights). Shadows cast by objects are automatically used to compute 3D depth, convexity, concavity, and distance.

• Semantic Regularities and Scene Schemas:

  • Semantic regularities: Typical functional characteristics, meanings, and activities associated with specific types of scenes.

  • Scene schema: A stored cognitive knowledge base specifying the objects, configurations, and spatial arrangements typically found within a given scene category (e.g., a kitchen typically contains a refrigerator, stove, and utensils).

  • Schema violations: When an object violates the semantic regularity of a setting (such as placing a kitchen-themed sign reading "Life is short lick the bowl" above a bathroom toilet), it causes cognitive conflict, delayed recognition, and altered perceptual interpretations.

Bayesian Inference in Perception

• The Bayesian Framework: Models perception as an optimal statistical inference process where the visual system evaluates sensory ambiguity by calculating the posterior probability of a physical world state given current sensory data.

• Mathematical Components of Bayesian Inference:

  • Prior Probability ($\text{Prior}$): The initial subjective probability or stored belief regarding the baseline frequency of a specific environmental hypothesis before new evidence is observed.

  • Likelihood: The probability that the received sensory input would occur under that specific hypothesis.

  • Posterior Probability ($\text{Posterior}$): The updated perceptual estimate, calculated by multiplying the prior by the likelihood:         $\text{Posterior} \propto \text{Prior} \times \text{Likelihood}$

• Diagnostic Case Example of Bayesian Inference:

  • Clinical symptom: An individual presents with an acute coughing symptom.

  • Candidate causes: Common cold, lung disease, or heartburn.

  • Prior probabilities: A person's stored medical base-rates indicate that colds are very frequent (high prior), heartburn is very frequent (high prior), and lung disease is rare (low prior).

  • Likelihoods: Known clinical associations indicate that both a cold and lung disease have a high likelihood of causing coughing, whereas heartburn has a very low likelihood of causing coughing.

  • Posterior calculation:

    • Cold: High prior $\times$ High likelihood $=$ Very high posterior probability.

    • Lung disease: Low prior $\times$ High likelihood $=$ Low to moderate posterior probability.

    • Heartburn: High prior $\times$ Low likelihood $=$ Very low posterior probability.

  • Perceptual / Diagnostic conclusion: The cough is inferred to be most likely caused by a cold.

The Perception-Action Cycle

• Reciprocal Coupling: Perception and action form an ongoing, circular feedback loop in daily life. Perceptual systems gather sensory data to guide physical movement, while active bodily movement generates new vantage points and sensory information that clarify perception.

• Movement Facilitating Perception: While moving through space introduces optical flow and visual motion blur, active movement provides multiple angular viewpoints, motion parallax, and depth cues that help the brain establish viewpoint invariance and construct stable 3D representations.

• Neural Communication via the $\mu$ (Mu) Rhythm: The primary somatosensory cortex and primary motor cortex communicate across the central fissure using a specialized neural oscillation termed the $\mu$ rhythm (operating in the $8\,Hz$ to $13\,Hz$ range). This oscillation synchronizes sensorimotor coordination and is distinct from the rhythms used by other sensory cortical areas.

Neuroanatomy and Dual Visual Processing Streams

• Cerebral Lobes and Functional Divisions:

  • Frontal lobe: Houses the primary motor cortex and motor association cortex; coordinates executive function, motor control, action knowledge, and higher-order planning.

  • Parietal lobe: Houses the primary somatosensory cortex and somatosensory association cortex; coordinates spatial attention, cross-modal integration, mental imagery, and spatial awareness.

  • Occipital lobe: Contains the primary visual cortex ($V1$) and early visual association cortices; dedicated to visual feature processing.

  • Temporal lobe: Contains the primary auditory cortex and specialized category-selective visual areas (for faces, bodies, objects, and scenes); supports long-term memory access, semantic categorization, and speech perception.

• Hierarchical Organization of Visual Processing:

  • Visual signals originate in the retina and project via the optic nerve through the lateral geniculate nucleus of the thalamus to the primary visual cortex ($V1$ / striate cortex).

  • Processing ascends hierarchically through secondary visual areas ($V2$, $V3$, $V4$, $V5$/MT) before diverging into two major, anatomically separate parallel processing streams:

• The Ventral Stream ("What" Pathway):

  • Trajectory: Extends from the occipital lobe ($V1$, $V2$, $V4$) inferiorly into the temporal lobe.

  • Functional specialization: Dedicated to object identification, shape analysis, color perception, and semantic categorization ("What is the object?").

• The Dorsal Stream ("Where" / "How" Pathway):

  • Trajectory: Extends from the occipital lobe ($V1$, $V2$, $V3$, $V5$/Area MT) superiorly into the parietal lobe.

  • Functional specialization: Dedicated to spatial localization, visual motion tracking, visually guided action, and motor interaction ("Where is the object?" and "How do I interact with it?").

• Neuropsychological Evidence: Double Dissociation:

  • Macaque Lesion Experiments (Ungerleider and Mishkin):

    • Temporal lobe ablation: Monkeys with bilateral lesions to the temporal cortex showed severe deficits in the Object Discrimination task (differentiating between a triangular prism and a rectangular block to obtain food), but remained completely unimpaired on the Landmark Discrimination task.

    • Parietal lobe ablation: Monkeys with bilateral lesions to the parietal cortex were severely impaired on the Landmark Discrimination task (choosing a food well located closest to a visual cylinder landmark), but showed intact performance on the Object Discrimination task.

    • Conclusion: Confirms a functional double dissociation where the temporal lobe mediates object recognition ("What") and the parietal lobe mediates spatial localization ("Where").

*   *Human Clinical Neuropsychology: Patient DF (Milner and Goodale):*
    *   Patient DF sustained bilateral damage to the ventral visual pathway (ventral stream) secondary to carbon monoxide poisoning, while her dorsal pathway and primary visual cortex ($V1$) remained intact.
    *   *Perceptual Orientation Matching task:* When shown a card and asked to orient it to match the angle of a tilted slot, DF performed at chance level, displaying random errors across orientations because she could not visually perceive or match object orientation.
    *   *Visuomotor Posting task:* When asked to physically insert the card into the slot (posting), DF moved her hand smoothly and rotated the card to the correct angle during the reach, matching control performance perfectly.
    *   *Conclusion:* Demonstrates that conscious visual perception of object features (the ventral "What" stream) is neurally dissociable from the visual control of motor action directed toward that object (the dorsal "Where/How" stream).