Perceptual Constancy
Perceptual constancy, also known as perceptual invariance, is the ability to recognize objects across variations in sensory input, such as a face from multiple angles, or a word spoken by different talkers1,2. Perceptual constancy requires that sensory systems, including vision and hearing, develop a level of tolerance to identity-preserving transformations3,4. In hearing, tolerance is critical for representing sounds such as individual words or phonemes across talkers, voice pitch, background noise and other acoustic transformations5, and is a key step in auditory object formation and scene analysis1,6,7.
perceptual constancy
Both humans and other animals perceive sound features constantly despite variation in sensory input: we can recognize loudness across variation in location8, frequency across sound level9 and sound identity across talkers10,11, vocal tract length12,13,14 and fundamental frequency (F0)15,16,17. At the neural level, tolerance is observed within auditory cortex, where neurons remain informative about the identity of vocalizations18,19,20, pure tones21 and pulse trains22 across variations in acoustic properties. For speech sounds such as vowels, multiple sound features including phoneme identity, location and F0 modulate activity of auditory cortical neurons23,24,25,26. However, tolerance has yet to be shown in subjects actively demonstrating perceptual constancy, and the behavioral relevance of previously demonstrated tolerant representations in auditory cortex remains unclear. Furthermore, although auditory cortical processing is modulated by attention and experience27, it is unknown how these processes affect tolerant representations.
Here, we asked if tolerant representations for complex sounds exist in early auditory cortex during perceptual constancy, how tolerance was related to behavior, and how tolerance was modulated by attention and experience. To address these questions, we recorded auditory cortical neurons in ferrets discriminating synthesized vowel sounds that varied across identity-preserving acoustic transformations including F0, sound location, level, and voicing. These features varied independently and thus represented orthogonal dimensions in feature space.
Have you ever noticed how snow looks just as "white" in the middle of the night under dim moonlight as it does during the day under the bright sun? When you walk away from an object, have you noticed how the object gets smaller in your visual field, yet you know that it actually has not changed in size? Thanks to perceptual constancy, we have stable perceptions of an object's qualities even under changing circumstances.
Perceptual constancy is the tendency to see familiar objects as having standard shape, size, color, or location, regardless of changes in the angle of perspective, distance, or lighting. The impression tends to conform to the object as it is assumed to be, rather than to the actual stimulus presented to the eye. Perceptual constancy is responsible for the ability to identify objects under various conditions by taking these conditions into account during mental reconstitution of the image.
Even though the retinal image of a receding automobile shrinks in size, a person with normal experience perceives the size of the object to remain constant. One of the most impressive features of perception is the tendency of objects to appear stable despite their continually changing features: we have stable perceptions despite unstable stimuli. Such matches between the object as it is perceived and the object as it is understood to actually exist are called perceptual constancies.
Within a certain range, people's perception of a particular object's size will not change, regardless of changes in distance or size change on the retina. The perception of the image is still based upon the actual size of the perceptual characteristics. The visual perception of size constancy has given rise to many optical illusions.
Our eyes aren't the only sensory organs that "trick" us into perceptual constancy. Our ears do the job as well. In music, we can identify a guitar as a guitar throughout a song, even when its timbre, pitch, loudness, or environment change. In speech perception, vowels and consonants are perceived as constant even if they sound very different due to the speaker's age, sex, or dialect. For example, the word "apple" sounds very different when a two year-old boy and a 30 year-old woman say it, because their voices are at different frequencies and their mouths form the word differently... but we perceive the sounds to be the same. This is thanks to auditory perceptual constancy!
Methodologically, two types of constancy were distinguished from each other: (a) constancy obtained by the comparative judgment in the two-stimulus-comparison situation (as has been studied traditionally) and (b) constancy obtained by the absolute judgment in the single-stimulus situation (as is the case in everyday experience). Some studies on the constancy of the latter type were reviewed.
The implications of the constancy were discussed in terms of (i) segregation of stimulation, (ii) perceptual constancy vs. thing constancy, (iii) comparative vs. absolute, and (iv) transient standards vs. lasting norms.
Location Constancy: This type of perceptual constancy refers to the connection between the object and the viewer. This is when an inanimate, stationary object appears stationary, despite any movement caused by the viewer either walking away or towards it.
Gestalt postulated the figure-ground rule. In perceptual constancy, this helps to explain how we separate familiar objects from foreign ones. When looking at a new scene, our perception tends to make familiar objects stand out, while strange objects fade into the background.
Perceptual constancy helps to explain why it is we see things in a certain way, regardless of how they are. It also helps to keep our external environment ordered and predictable, helping us to feel a sense of safety.
In this video I describe perceptual constancy, which refers to the idea that we perceive a relatively stable and unchanging world despite the fact that sensory information is changing dramatically. I explain how this applies visually to size, brightness, and shape. We often forget this learning process because it occurred during infancy, but it is demonstrated clearly in previously blind individuals who must slowly learn how to see following eye surgery.
The tendency for a perceived object or a perceptual quality to appear the same when the pattern of sensory stimulation or proximal stimulus alters through a change in orientation, distance, illumination, or some other extraneous influence or factor. See colour constancy, lightness constancy, melodic constancy, object constancy, odour constancy, person constancy, position constancy, shape constancy, size constancy, velocity constancy, word constancy. See also Brunswik ratio, Land effect, phenomenal regression, retinex theory, shape-slant invariance, template matching, Thouless ratio.
noun. ['ˈkɑːnstənsi'] (psychology) the tendency for perceived objects to give rise to very similar perceptual experiences in spite of wide variations in the conditions of observation.
This project has as its focus a pair of related phenomena central to human perception. The first is the underdetermination of perceptual content by sensor input, and the second is a class of mechanisms designed to transform impoverished sensor input into useful perceptual content, mechanisms commonly called `perceptual constancies'. The goal of this project is to discuss a particularly difficult form of sensory underdetermination I call \textitstacking, a \textitco-local sensory conflation of distal properties like surface color and illumination, or size and distance. And although stacking problems are not computationally intractable, there appear to be significant constraints on potential solutions to these problems, constraints rooted in the phenomenal structure of perceptual experience. Accordingly, I spend much of the project examining in detail a type of intentional content I argue explains the phenomenology of perceptual experience---\textitphenomenal content. With explanatory adequacy as a guiding principle, I argue that phenomenal contents cannot be indeterminate, in the sense that they cannot fail to be attributional. I also argue that phenomenal contents must be \textitphenomenally bounded, a conclusion that rules out of consideration various high-level properties like natural kinds. Additionally, I argue that there are in fact two distinct domains of phenomenal content. The first I call the \textitenergy map, as its contents are tied closely to the distribution of energy across the sensors. The second I call the \textitworldmaker, as its contents are what make the world accessible and intelligible to perceivers. Worldmaker contents are underdetermined by energy map contents and are a phenomenal manifestation of perceptual mechanisms of \textitoverreach, the general term for solutions to problems of sensory underdetermination. After setting up these constraints, I apply them to the problem of color constancy, and I argue that the standard view of color constancy, found both in philosophy and in perceptual psychology---the view that we perceptually represent spectral surface reflectance---must be wrong. My methodology and conclusions have implications both for philosophical accounts of perception and for computational models of perception.
This project has as its focus a pair of related phenomena central to human perception. The first is the underdetermination of perceptual content by sensor input, and the second is a class of mechanisms designed to transform impoverished sensor input into useful perceptual content, mechanisms commonly called `perceptual constancies'. The goal of this project is to discuss a particularly difficult form of sensory underdetermination I call \\textitstacking, a \\textitco-local sensory conflation of distal properties like surface color and illumination, or size and distance. And although stacking problems are not computationally intractable, there appear to be significant constraints on potential solutions to these problems, constraints rooted in the phenomenal structure of perceptual experience. Accordingly, I spend much of the project examining in detail a type of intentional content I argue explains the phenomenology of perceptual experience---\\textitphenomenal content. With explanatory adequacy as a guiding principle, I argue that phenomenal contents cannot be indeterminate, in the sense that they cannot fail to be attributional. I also argue that phenomenal contents must be \\textitphenomenally bounded, a conclusion that rules out of consideration various high-level properties like natural kinds. Additionally, I argue that there are in fact two distinct domains of phenomenal content. The first I call the \\textitenergy map, as its contents are tied closely to the distribution of energy across the sensors. The second I call the \\textitworldmaker, as its contents are what make the world accessible and intelligible to perceivers. Worldmaker contents are underdetermined by energy map contents and are a phenomenal manifestation of perceptual mechanisms of \\textitoverreach, the general term for solutions to problems of sensory underdetermination. After setting up these constraints, I apply them to the problem of color constancy, and I argue that the standard view of color constancy, found both in philosophy and in perceptual psychology---the view that we perceptually represent spectral surface reflectance---must be wrong. My methodology and conclusions have implications both for philosophical accounts of perception and for computational models of perception. 041b061a72