The kind of stimulus used by Navon (1977) to demonstrate the importance of global features in perception.

as the one shown in Figure 4.2. In one of his studies, participants had to decide as rapidly as possible on some trials whether the large letter was an "H" or an "S"; on other trials, they had to decide whether the small letters were Hs or Ss. Performance speed with the small letters was greatly slowed when the large letter was different from the small letters. In contrast, decision speed with the large letter was unaffected by the nature of the small letters. According to Navon (1977, p. 354), these findings indicate that, "perceptual processes are temporally organised so that they proceed from global structuring towards more and more fine-grained analysis. In other words, a scene is decomposed rather than built up."

Some of the available evidence is inconsistent with Navon's conclusion. Kinchla and Wolfe (1979) used stimuli of a similar nature to those of Navon (1977), but of variable size. When the large letter was very large, processing of the small letters preceded processing of the large letter. They argued that global processing occurs prior to more detailed processing only when the global structure of a pattern or object can be ascertained by a single eye fixation.

The main problem with research stemming from Navon's (1977) study is that it has not proved possible to identify precisely where in the visual processing system the global advantage occurs. In the words of Kimchi (1992, p. 36):

There seems to be evidence, though not entirely conclusive, that global advantage occurs at early perceptual processing. Certain findings suggest that the mechanisms underlying the effect may be sensory, but other findings are suggestive of attentional mechanisms.

Cognitive neuroscience

Cognitive neuroscientists have obtained evidence of some relevance to feature theories. If the presentation of a visual stimulus leads initially to detailed processing of its basic features, then we might be able to identify cells in the cortex involved in such processing. However, the existence of cells specialised for responding to specific aspects of visual stimuli may be consistent with feature theories, but does not demonstrate that they are correct.

Hubel and Wiesel (e.g., 1979), used single-unit recordings to study individual neurons (see Chapter 1). They found that many cells responded in two different ways to a spot of light depending on which part of the cell was affected:

1. An "on" response, with an increased rate of firing while the light was on.

2. An "off' response, with the light causing a decreased rate of firing.

Many retinal ganglion cells, lateral geniculate cells, and layer IV primary visual cortex cells can be divided into on-centre cells and off-centre cells. On-centre cells produce the on-response to a light in the centre of their receptive field and an off-response to a light in the periphery; the opposite is the case with off-centre cells.

Hubel and Wiesel (e.g., 1979) discovered the existence of two types of neurons in the receptive fields of the primary visual cortex: simple cells and complex cells. Simple cells have "on" and "off" regions, with each region being rectangular in shape. Simple cells play an important role in detection. They respond most to dark bars in a light field, light bars in a dark field, or to straight edges between areas of light and dark. Any given simple cell only responds strongly to stimuli of a particular orientation, and so the responses of these cells could be relevant to feature detection.

There are many more complex cells than simple cells. They resemble simple cells in that they respond maximally to straight-line stimuli in a particular orientation. However, there are significant differences:

1. Complex cells have larger receptive fields.

2. The rate of firing of a complex cell to any given stimulus depends very little on its position within the cell's receptive field; in contrast, simple cells are divided into "on" and "off' regions.

3. Most complex cells respond well to moving contours, whereas simple cells respond only to stationary or slowly moving contours.

There is also evidence for the existence of hypercomplex cells. These cells respond most to rather more complex patterns than do simple or complex cells. For example, some respond maximally to corners, whereas others repond to other various specific angles.

It is important to note that cortical cells provide ambiguous information, because they respond in the same way to different stimuli. For example, a cell that responds maximally to a horizontal line moving slowly may respond moderately to a horizontal line moving rapidly and to a nearly horizontal line moving slowly. Thus, as Sekuler and Blake (1994, p. 134) pointed out, "Neurons in the visual cortex cannot really be called 'feature detectors',...because individual cells cannot signal the presence of a particular visual feature with certainty."

Hubel and Wiesel (1962) argued that processing in the visual cortex is based on straight lines and edges. An alternative view is based on gratings, which are patterns consisting of alternating lighter and darker bars. Of particular importance are sinusoidal gratings, in which there are gradual intensity changes between adjacent bars. According to Sekuler and Blake (1994), gratings possess four properties:

1. Spatial frequency: the spacing of bars as imaged on the retina.

2. Contrast: the difference in intensity of light and dark bars.

3. Orientation: the angle at which the bars of the grating are presented.

4. Spatial phase: the position of the grating with respect to some landmark (e.g., edge of a display).

It is possible to construct any desired visual pattern by manipulating each of these four properties of gratings.

Campbell and Robson (1968) assumed that the visual system contains sets of neurons (or channels) that respond to different spatial frequencies of gratings, and this assumption formed the basis of their multichannel model. They obtained some support for their model by presenting people with compound gratings, which were formed by combining a number of simple sinusoidal gratings. The visual system responded differently to each of the components of these compound gratings, presumably because the channels appropriate to each component were being activated. Subsequent research indicated that most cells in the primary visual cortex respond more strongly to sinusoidal gratings than to lines and edges (see Pinel, 1997).

The emphasis on spatial frequency led to the development of the contrast sensitivity function, which indicates an individual's ability to detect targets of various spatial frequencies. Evidence that the contrast sensitivity function is a valuable measure was reported by Ginsburg, Evans, Sekuler, and Harp (1982). Pilots flew simulated missions in an aircraft simulator in conditions of reduced visibility, and sometimes had to abort a landing because the runway was blocked. Ginsburg et al. (1982) assessed the pilots' visual acuity, which is an assessment of the smallest detail that can be detected. The pilots' flying performance was not related to visual acuity. However, those pilots with the highest contrast sensitivities noticed that the runway was blocked from a greater distance than did those with the lowest contrast sensitivities.

Harvey, Roberts, and Gervais (1983) presented individual letters very briefly, and asked their participants to name them. Some letters (e.g., "K" and "N") having several features in common were not confused, which is contrary to the prediction of feature theory In contrast, letters with similar spatial frequencies tended to be confused, even if they shared few common features. These findings suggest that spatial frequency is more important than features in the representation of letters within the visual system.


Stimulus features play a role in pattern recognition. However, feature theories leave much that is of importance out of account. First, they de-emphasise the effects of context and of expectations on pattern recognition. Weisstein and Harris (1974) used a task involving detection of a line embedded either in a briefly flashed three-dimensional form or in a less coherent form. According to feature theorists, the target line should always activate the same feature detectors, and so the coherence of the form in which it is embedded should not affect detection. In fact, target detection was best when the target line was part of a three-dimensional form. Weisstein and Harris (1974) called this the "object-superiority effect", and this effect is inconsistent with many feature theories.

Second, pattern recognition does not depend solely on listing the features of a stimulus. For example, the letter "A" consists of two oblique uprights and a dash, but these three features can be presented in such a way that they are not perceived as an A: \ / —. In order to understand pattern recognition, we need to consider the relationships among features as well as simply the features themselves.

Third, the limitations of feature theories are clearer with three-dimensional than with two-dimensional stimuli. The fact that observers can generally recognise three-dimensional objects even when some of the major features are hidden from view is hard to account for on a theory that emphasises the role of features in recognition.

Fourth, global processing often precedes feature processing (e.g., Navon, 1977). Additional evidence comes from research on face processing (discussed later).

Stop Anxiety Attacks

Stop Anxiety Attacks

Here's How You Could End Anxiety and Panic Attacks For Good Prevent Anxiety in Your Golden Years Without Harmful Prescription Drugs. If You Give Me 15 minutes, I Will Show You a Breakthrough That Will Change The Way You Think About Anxiety and Panic Attacks Forever! If you are still suffering because your doctor can't help you, here's some great news...!

Get My Free Ebook

Post a comment