PSY380: VISION SCIENCE
1) Questions: - Who are you and why are you here? (Why vision?) - What is visual perception? - What is the function of visual perception? 2) The syllabus & instructor 3) Lecture 1
What is visual perception? Visual perception is the process of acquiring knowledge about environmental objects and events by extracting information from the light they emit or reflect (Palmer p5). A visual system is a set of interacting parts that 1) responds to light (thus extracting information), 2) integrates the information in a way that can be used to guide behavioral processes.
Function & form Image from Gomez-Marin & Louis, Curr. Opinion Neurobiology, 2012 24 light sensors allow fly larvae to move away from light and toward moving food sources
Function & form Vertebrate typically have two eyes, but their visual systems are developed for different needs
Function & form NASA 10 cameras and other light-based sensors enable Curiosity to navigate, analyze, and relay information to humans
Lecture 1: Top-down
Terms and concepts Part 0: visual perception, visual systems Part 1: emission vs. intromission, photons, unconscious inference, underdetermined (or underconstrained), the inverse problem, ontogenetic and phylogenetic, structuralism, gestaltism, emergent property, psychophysiological isomorphism, physical gestalt, attractor, ecological optics, optic flow, constructivism, likelihood principle, top-down versus bottom-up processing Part 2: levels of analysis, mechanism, central nervous system, peripheral nervous system, neurons, dendrite, axon, excitatory neurons, inhibitory neurons, interneurons, projection neurons, hierarchy of reflex arcs
A brief history of vision (why?)
Classical theories of vision: emission versus intromission Emission: light shines out of the eye to reveal the world around us Intromission: light comes from the environment and acts on the eye - 5 th century BCE: Empedocles proposes that fire shines out of the eye. - 300 BCE: Euclid suggests light travels infinitely fast to make this work. - 2 nd century CE: Ptolemy and Galen both endorse the theory - 11 th century: Ibn al-haytham (Alhazen), the first optical physicist, performed experiments yielding results consistent with intromission
Flash forward to present-day physics (Lecture 2) - Light is carried by units/waves of energy called photons - Released when something (sun, light bulb) expels energy - More energy higher wave frequency - Photons always travel at the same speed (299,792 km/s) Inductiveload, NASA
Immanuel Kant, Critique of Pure Reason (1781) We don t just take-in the world. We have preset structures (including structures for time and space) that interact with the world to generate our perception.
Helmholtz (1821-1894) Unconscious inference: there is a gap between what information is available to us and the amount of information we can get from it. (current terminology: visual information is underdetermined or underconstrained: the inverse problem) Our minds bridge this gap by making inferences using hidden assumptions about the world. These assumptions are built from ontogenetic or phylogenetic learning
Example of how our unconscious inferences can be used against us (images from: https://www.youtube.com/watc h?v=zzd0ad0cxmg)
Structuralism (late 19 th, early 20 th century) Believed that perceptions are made-up of sensory atoms. The links between sensory atoms make-up associations, and therefore memories. Links are created by repeated, coincident activity Wilhelm Wundt (1832-1920) in his psych lab
Fast forward to 1948 (and lecture 6: learning & memory) Donald Hebb, The Organization of Behavior (1948) Links are created by repeated, coincident activity Led to Hebb s theory of cell assemblies
Gestaltism Stimulus configurations have emergent properties Emergent property: a pattern that arises from interactions among smaller or simpler entities that don t themselves exhibit the property. Examples of configurations that are more than the sum of their parts: Reification: we sometimes construct what isn t there (e.g., Kanizsa triangle) Multistability: we do not simultaneously perceive the same image to mean different things (e.g., Necker cube) Invariance: we can perceive very different sensory stimuli as being the same thing
(R. Beau Lotto & Dale Purves, Nature Neuroscience, 1999)
(R. Beau Lotto & Dale Purves, Nature Neuroscience, 1999)
Some Gestaltists and their ideas about the brain Max Wertheimer Wolfgang Kohler Believed in psychophysiological isomorphism: perceptual experiences are isomorphic with underlying brain events Kohler believed that the brain was a physical Gestalt A dynamic physical system that converged toward an equilibrium state of minimal energy
Fast forward to 1980 s-present (and Lecture 2) In networks of nodes (say, neurons) that can influence each other, the set or pattern of nodes that are active often converges (or settles) into low energy states called attractors (Eliasmith, Scholarpedia) (Rolls & Treves, Neural Networks and Brain Function, 1998)
James J. Gibson & Ecological Optics: when taking into account features such as texture and optic flow (movement of stimulus relative to eye), most stimuli are not underdetermined
Texture influencing distance perception
Motion parallax influencing distance perception
Constructivism Ideas can be traced back to Helmholtz: the brain constructs our perception based on assumptions built from ontogenetic and phylogenetic learning Likelihood principle: our perceptions are largely based on hypotheses about what we might be or should be seeing, based on combining context with information transmitted from the retina. Core idea underlying constructivism: both topdown and bottom-up processes contribute to visual perception
Important concept list 1) Both top-down and bottom-up processes contribute to visual perception What might this mean? Let s also consider nervous system structure
Part 2: Overview of the nervous system and its computation
The simplest circuit in the nervous system (albeit, not a visual circuit):
How can we explain a reflex? 1) What does the system do? Transforms a specific sensory input into a specific motor output according to specific rules This is the system s behavior or computation
How can we explain a reflex? 2) What is the system made of? Cells, including neurons, a substrate that includes electrically charged particles. The neurons themselves exhibit certain behaviors, including action potentials that change in frequency over time (firing rates). This is the system s hardware or material.
How can we explain a reflex? 3) What do the material parts of the system do to accomplish what the system does? The environment stimulates one neuron to increase firing rates (according to rules dictated by molecular actions and interactions), this stimulates a second (according to a new set of rules) that in turn stimulates a muscle to contract. This is the system s algorithm. Alternatively: the mechanism
A mechanism is a description of a system s behavior built from a description of the actions and interactions of the system s parts.
Levels of analysis Psychology and systems neuroscience (this class) ecosystems Ecology, Social Sciences, Economics groups/social systems organisms/ individual animals networks of circuits neural circuits Physics, Chemistry cells/neurons Fundamental Particles/forces atoms molecules organelles
Back to our question: What does it mean for a process to be both topdown and bottom-up?
Organizing principles of the nervous system The nervous system is organized in a hierarchy of reflex arcs (i.e., sensorymotor/perception-action loops) Charles Sherrington The integrative actions of the nervous system, 1906 Sensory Environment Motor
Organizing principles of the nervous system The nervous system is organized in a hierarchy of reflex arcs (i.e., sensorymotor/perception-action loops) Charles Sherrington The integrative actions of the nervous system, 1906 Extracting information according to expectations and relevance Integrated sensory Sensory Motor Action Environment
Organizing principles of the nervous system The nervous system is organized in a hierarchy of reflex arcs (i.e., sensorymotor/perception-action loops) Higher-order perceptions Action plans Charles Sherrington The integrative actions of the nervous system, 1906 Extracting information according to expectations and relevance Integrated sensory Sensory Environment Motor Action
Organizing principles of the nervous system The nervous system is organized in a hierarchy of reflex arcs (i.e., sensorymotor/perception-action loops) Charles Sherrington The integrative actions of the nervous system, 1906 Learned associations & relevance influence information extraction Higher-order perceptions Top-down Bottom-up Integrated sensory Sensory Environment Motor Action plans Action
brain region Simplified (largely over-simplified) diagram depicting topdown and bottom-up processes Hippocampal formation Environmental context Temporal cortex Object identity & associations information Occipital Cortex Visual features
Research can also be top-down and bottom-up: Psychology and systems neuroscience (this class) ecosystems Ecology, Social Sciences, Economics Physics, Chemistry groups/social systems organisms/ individual animals Top-down Bottom-up networks of circuits neural circuits cells/neurons Fundamental Particles/forces atoms molecules organelles
Summary: there are different ways in which the term top-down can be used - Higher level of information abstraction (or integration) acts on a lower level to generate our visual perception - Observations at different levels of complexity (or embedded systems) inform hypotheses about the mechanisms by which systems behave.
Preview to lecture 2: an 8 ½ min video on how the human visual system works (specifically 1:42-7:40) https://www.youtube.com/watch?v=aulr0kzfwbu