Chapter 1: General principles

General principles of sensation and perception (pp. 12ó23)

Ask Yourself

What you need to know

  1. Physiological Principles (pp. 12ó16)
    • Specific nerve energy
    • Neuron physiology and hierarchical pathways
    • Selectivity
    • Organisation
    • Cortical magnification
    • Plasticity
    • Noise
  2. Perceptual Principles (pp. 17ó19)
    • Sensations and qualia
    • Detection
    • Stevens's power law
    • Adaptation
  3. Theoretical Principles (pp. 19ó23)
    • Representations: analogue and symbolic
    • Computations and algorithms
    • Linking propositions
    • Decision rules

Physiological Principles

Sensory receptor cells must be specialised to transduce different forms of environmental energy into common electrical signals. The breakdown of light-sensitive pigment within photoreceptors, and the vibrations of cilia in mechanoreceptors, both lead to neural impulses. Nerve impulses cannot differentiate between the senses; the cortical destination of the impulses determines the sensory experience (visual or auditory). Muller (1838; see FP p. 15) first formalised this concept in the law of specific nerve energy, which gains empirical support from electrical stimulation of the cortex in awake, anaesthetised patients.

Electrical signals travel from a cell's dendrites, via its axon, to terminal buttons that synapse with the dendrites of receiving neurons. Signals are transmitted across synapses by either excitatory or inhibitory neurotransmitters (e.g. acetylcholine and gamma amino butyric acid, respectively). Between sensory receptor cells and cortical receiving areas, there are three synapses in the visual sensory pathway, and five in the auditory sensory pathway.

In all of the senses, except olfaction, receptor signals travel unidirectionally to the thalamus. From the thalamus the flow of information is bi-directional, to and from the primary cortex (receiving area), and subsequently to and from the secondary cortex (cortical association area). Each successive stage in the processing hierarchy refines and modifies the sensory signal.

Each sensory system responds only to a particular range of stimuli. For example, the human auditory system responds to sound pressure waves of frequencies between 20 Hz and 16,000 Hz.

All sensory systems respond to a selective range of stimuli that define their sensory space. Within this range, individual neurons have their own, highly selective stimulus preferences (Fig 1.11 shows the preferences for a cat's visual cortical cell; see FP p. 15). The area in which a preferred stimulus must be presented to elicit a response is known as the cell's receptive field.

At each level of sensory processing, neighbouring cells generally have similar stimulus preferences. Tootell et al. (1982; see FP p. 14) used a physiological staining technique to identify active cells in cortex. Their illustration of the topographic map of the visual field in the striate cortex provides a striking example of cells being organised according to proximity of their receptive fields. Cortical maps are not necessarily scale representations. Cortical magnification, or exaggeration of part of a sensory dimension, often occurs.

Plasticity, or adaptability, is a feature of all sensory processing. Neural mechanisms can be modified to accommodate either short-term environmental changes, or long-term developmental changes.

The activity level in neurons typically varies between 0 and 200 impulses per second. Sensory neurons are subject to noise: random variations in firing frequency that are unrelated to environmental stimulation. The random fluctuations that occur in the permeability of a neuron's ion channels, and in the chemical reactions across its synapse, have been identified by White et al. (2000; see FP p. 16) as two potential sources of noise.

Perceptual Principles

Sensations are private, conscious mental states, generally thought to correspond to particular physical brain states. Qualitative attributes of a sensation, such as loudness or colour, are known as qualia. How particular patterns of neural activity evoke sensory experiences remains unexplained.

Stimulus detection is a probabilistic event, increasing with stimulus level according to a standard "S-shaped" psychometric function. The response transition from no detection to detection is gradual, rather than abrupt, and almost certainly reflects the influence of neural noise.

The technique of magnitude estimation relates the physical magnitude of a stimulus to its perceived magnitude. The relationship is non-linear, conforming to Stevens's power law (1961; see FP p. 18), according to which:

All sensory systems adapt to provide the best match between their limited sensory space and the external stimulation. After adaptation to a relatively intense stimulus:

Theoretical Principles

Perceptual experiences create particular patterns of neural activity, which are representations of the external world, at every level of sensory processing, from receptors to secondary cortex.

As a general concept, there are two forms of representation:

  1. Analogue representationsórelate two systems using proportional values such as spatial position or response rate (e.g. a visual illustration of a particular species of bird).
  2. Symbolic representationsórelate two systems using discrete symbols such as words or characters (e.g. a verbal description of a particular species of bird).

As a perceptual concept, there are two forms of representations in the brain:

  1. Analogueóbrain states are directly and proportionately related to the external stimulation (e.g. topographic maps in the visual cortex).
  2. Symbolicóbrain states are abstract symbols signifying perceptual objects or object properties (e.g. "umbrella", "bird", "smooth").

The nature of a representation determines the type of computation that can be performed. Algorithms that apply to:

  1. Analogue representations describe how quantities are manipulated mathematically.
  2. Symbolic representations describe how symbols are compared, combined, and created.

Since analogue computations capture the properties of many cortical cells, low-level perceptual representations are usually associated with analogue representations. High-level perceptual representations are generally regarded as symbolic.

Table 1.1: Summary of analogue and symbolic representations
Representation General concept Perceptual concept Type of computation Neural processes
Analogue Proportional values used to relate two systems Brain states proportionately related to external stimulation Quantities manipulated mathematically (signal processing) Low-level perceptual, e.g. population coding
Symbolic Discrete symbols used to relate two systems Brain states symbolise external stimulation Symbols compared, combined, and created High-level perceptual and cognitive, e.g. object identity, problem-solving

Rigorous theories of perception:

So What Does This Mean?

Following transduction, the information carried in nerve impulses is repeatedly refined as it progresses through successive stages of hierarchical analysis from receptors to higher cortex. At every level, neurons are organised according to their individual, highly specific stimulus preferences. The cortical destination of nerve impulses differentiates the experiences of different senses.

There is a reliable relationship between the physical magnitude of a stimulus and its probability of detection and, for a detectable stimulus, perceived magnitude is reliably related to physical magnitude, according to Stevens's power law.

Many properties of low-level perceptual neurons are captured by analogue representations and computations. Higher-level perceptual processing is generally regarded as symbolic.

Other topics in this chapter