3.2 Selectivity in the Visual System

As I showed in the fi rst chapter, not every piece of data is relevant or useful

to a system or process in an organism. Thus, I posited data selectivity as

the property of a subsystem or process that allows for discrimination

between relevant and irrelevant data. Data selectivity is necessary so that

the raw data actually can become informative for a subsystem or process.

Data that have been selected cease to be of the kind that are potentially

useful and become actual information. It should be clear that data selectivity

is integral to the processing of information in the subsystems and

processes of a hierarchically organized system. However, as I went on to

note, the processing of information also requires fl exible communication

between afferent and efferent entities, in an environment, so that a change

can be evoked in the activity of the efferent entity. Finally, some kind of

storage or imprinting mechanism would have to exist so that the information

can be infl uential for the efferent entity.

Now, in the very simplest of terms, the complex processing of information

in the nervous system seems to require the fourfold steps of (1) detecting

data in some environment, (2) discriminating between relevant and

irrelevant data, (3) integrating information, and (4) initiating some kind

of response, as Audesirk et al. (2002), Sekuler & Blake (2002), and Kandel

et al. (2000) each have noted in their own ways. The goal of this section

is to focus upon the second step of this process, namely, discriminating

between relevant and irrelevant data. I will demonstrate that there is data

selectivity occurring at virtually every level of visual processing, from the

activities of photoreceptors in the retina, to the columnar and blob-cell

fi rings, all the way up through the what and where unimodal systems, to

the unimodal and multimodal association processes occurring in the occiptotemporal,

parietotemporal, prefrontal, and limbic cortices. This data

selectivity makes it possible for the components of the visual system to

process data and make use of this data as information.

Visual processing is already occurring in the retina, and this entails that

the various kinds of neurons therein actively are selecting data that are

relevant to their specifi c function. The retinal cells are specialized to detect

differences in the intensity of light falling upon them, since the rods selectively

attend to dim light, while the cones selectively attend to intense

light. Sekuler & Blake (2002, p. 91) underscore this selective capacity in the retina by noting that “events in the retina shape vision by emphasizing

some information and by de-emphasizing other information.”

This information from the retina then is sent through the ganglion

cells to the brain. Data selectivity continues in the occipital lobe where

orientation-specifi c and ocular-dominance columns respond to lines and

depth, respectively, while the blobs process wavelength information, which

ultimately contributes to the recognition of various colors of objects. The

specifi cation of functions in the M and P cell pathways further attests to

this data selective property of the visual system, since the M cells respond

to depth, motion, and object position, while the P cells respond to form

and color.

The most complex level of the visual system makes connections in the

multimodal association areas of the prefrontal, parietotemporal, and

limbic cortices. Research shows that the prefrontal areas primarily are

responsible for motor planning, judgment, some memory, and language

production, while the limbic cortices are responsible for olfaction,

emotion, and some memory formation as well. Research also shows

that the parietotemporal lobe aids in sensory integration of visual space

and language but, most importantly, spatial attention (Lux, Marshall,

Ritzl, Zilles, & Fink, 2003; Milner & Goodale, 1995; Wurtz, Goldberg, &

Robinson, 1982).

Treisman (1977, 1988), Julesz (1983, 1984), and Desimone & Duncan

(1995) have proposed a mechanism of attention whereby the brain selectively

associates the disparate features of the visual scene for a short

time. The associated data are considered as spotlighted and comprise the

coherent visual scene of which an animal is aware (cf. Wurtz, Goldberg,

& Robinson, 1982; Posner & Peterson, 1990; Posner & Dahaene, 1994).

Treisman (1988) has performed psychological tests on humans demonstrating

this association, while tests on macaque monkeys—as well as

PET and fMRI tests on humans—reveal areas of the occiptotemporal

cortex to be active while subjects try to make a visual scene coherent

(Brefczynski & DeYoe, 1999; DeYoe, Felleman, van Essen, & McClendon,

1994; DeYoe, Trusk, & Wong-Riley, 1995; Beason-Held et al., 1998; Buckner

et al., 1995; Honda, Wise, Weeks, Deibel, & Hallett, 1998; Rushworth,

Daus, & Sipila, 2001).

The spotlight metaphor is helpful, since the data from experiments

suggest that neuronal processes ignore or discard nonuseful data while

selectively attending to or spotlighting relevant data. This is why Kandel et al.

(2000, p. 504) can maintain that “selective attention acts to limit the

amount of information that reaches the highest centers of processing in

The Visual System 69

the brain,” and Bear et al. (2001, p. 569) claim that attention “has to do

with preferential [italics mine] processing of sensory information” (also see

Sekuler & Blake, 2002; Fink et al., 1996).

Animals are bombarded with sensory data in droves. There would be no

way for the sensory systems of the animal to take in all of this data; if they

did, the animal probably would cease to function altogether, much like an

overloaded computer that shuts down (cf. Johnson-Laird, 1988). Thus,

there are selectivity mechanisms—kinds of fi ltering devices—that exist at

the various levels in the visual hierarchy, segregating relevant from irrelevant

data. The relevant data become processed as information, while the

irrelevant data are simply ignored. As was communicated to me by Charles

Anderson at a neuroscience conference at Washington University in St.

Louis, the visual system exhibits its own checkpoints of selectivity, from

the interactions among organelles in the neuron, to the retina’s ability to

detect differences in the intensity of light, to the spotlighting of visual

information at the higher levels of the visual hierarchy. This is why, while

researching the visual system, Anderson and his colleagues (van Essen

et al., 1994, p. 271) call our attention to mechanisms “for dynamically

regulating [italics mine] the fl ow of information within and between cortical

areas.” Also, as Zigmond et al. (1999, p. 822) note, in the visual system

“high-level neurons classify visual stimuli by integrating information that

is present in the earlier stages of processing, but also by ignoring information

that is independent of that classifi cation.”

As I showed in the fi rst chapter, not every piece of data is relevant or useful

to a system or process in an organism. Thus, I posited data selectivity as

the property of a subsystem or process that allows for discrimination

between relevant and irrelevant data. Data selectivity is necessary so that

the raw data actually can become informative for a subsystem or process.

Data that have been selected cease to be of the kind that are potentially

useful and become actual information. It should be clear that data selectivity

is integral to the processing of information in the subsystems and

processes of a hierarchically organized system. However, as I went on to

note, the processing of information also requires fl exible communication

between afferent and efferent entities, in an environment, so that a change

can be evoked in the activity of the efferent entity. Finally, some kind of

storage or imprinting mechanism would have to exist so that the information

can be infl uential for the efferent entity.

Now, in the very simplest of terms, the complex processing of information

in the nervous system seems to require the fourfold steps of (1) detecting

data in some environment, (2) discriminating between relevant and

irrelevant data, (3) integrating information, and (4) initiating some kind

of response, as Audesirk et al. (2002), Sekuler & Blake (2002), and Kandel

et al. (2000) each have noted in their own ways. The goal of this section

is to focus upon the second step of this process, namely, discriminating

between relevant and irrelevant data. I will demonstrate that there is data

selectivity occurring at virtually every level of visual processing, from the

activities of photoreceptors in the retina, to the columnar and blob-cell

fi rings, all the way up through the what and where unimodal systems, to

the unimodal and multimodal association processes occurring in the occiptotemporal,

parietotemporal, prefrontal, and limbic cortices. This data

selectivity makes it possible for the components of the visual system to

process data and make use of this data as information.

Visual processing is already occurring in the retina, and this entails that

the various kinds of neurons therein actively are selecting data that are

relevant to their specifi c function. The retinal cells are specialized to detect

differences in the intensity of light falling upon them, since the rods selectively

attend to dim light, while the cones selectively attend to intense

light. Sekuler & Blake (2002, p. 91) underscore this selective capacity in the retina by noting that “events in the retina shape vision by emphasizing

some information and by de-emphasizing other information.”

This information from the retina then is sent through the ganglion

cells to the brain. Data selectivity continues in the occipital lobe where

orientation-specifi c and ocular-dominance columns respond to lines and

depth, respectively, while the blobs process wavelength information, which

ultimately contributes to the recognition of various colors of objects. The

specifi cation of functions in the M and P cell pathways further attests to

this data selective property of the visual system, since the M cells respond

to depth, motion, and object position, while the P cells respond to form

and color.

The most complex level of the visual system makes connections in the

multimodal association areas of the prefrontal, parietotemporal, and

limbic cortices. Research shows that the prefrontal areas primarily are

responsible for motor planning, judgment, some memory, and language

production, while the limbic cortices are responsible for olfaction,

emotion, and some memory formation as well. Research also shows

that the parietotemporal lobe aids in sensory integration of visual space

and language but, most importantly, spatial attention (Lux, Marshall,

Ritzl, Zilles, & Fink, 2003; Milner & Goodale, 1995; Wurtz, Goldberg, &

Robinson, 1982).

Treisman (1977, 1988), Julesz (1983, 1984), and Desimone & Duncan

(1995) have proposed a mechanism of attention whereby the brain selectively

associates the disparate features of the visual scene for a short

time. The associated data are considered as spotlighted and comprise the

coherent visual scene of which an animal is aware (cf. Wurtz, Goldberg,

& Robinson, 1982; Posner & Peterson, 1990; Posner & Dahaene, 1994).

Treisman (1988) has performed psychological tests on humans demonstrating

this association, while tests on macaque monkeys—as well as

PET and fMRI tests on humans—reveal areas of the occiptotemporal

cortex to be active while subjects try to make a visual scene coherent

(Brefczynski & DeYoe, 1999; DeYoe, Felleman, van Essen, & McClendon,

1994; DeYoe, Trusk, & Wong-Riley, 1995; Beason-Held et al., 1998; Buckner

et al., 1995; Honda, Wise, Weeks, Deibel, & Hallett, 1998; Rushworth,

Daus, & Sipila, 2001).

The spotlight metaphor is helpful, since the data from experiments

suggest that neuronal processes ignore or discard nonuseful data while

selectively attending to or spotlighting relevant data. This is why Kandel et al.

(2000, p. 504) can maintain that “selective attention acts to limit the

amount of information that reaches the highest centers of processing in

The Visual System 69

the brain,” and Bear et al. (2001, p. 569) claim that attention “has to do

with preferential [italics mine] processing of sensory information” (also see

Sekuler & Blake, 2002; Fink et al., 1996).

Animals are bombarded with sensory data in droves. There would be no

way for the sensory systems of the animal to take in all of this data; if they

did, the animal probably would cease to function altogether, much like an

overloaded computer that shuts down (cf. Johnson-Laird, 1988). Thus,

there are selectivity mechanisms—kinds of fi ltering devices—that exist at

the various levels in the visual hierarchy, segregating relevant from irrelevant

data. The relevant data become processed as information, while the

irrelevant data are simply ignored. As was communicated to me by Charles

Anderson at a neuroscience conference at Washington University in St.

Louis, the visual system exhibits its own checkpoints of selectivity, from

the interactions among organelles in the neuron, to the retina’s ability to

detect differences in the intensity of light, to the spotlighting of visual

information at the higher levels of the visual hierarchy. This is why, while

researching the visual system, Anderson and his colleagues (van Essen

et al., 1994, p. 271) call our attention to mechanisms “for dynamically

regulating [italics mine] the fl ow of information within and between cortical

areas.” Also, as Zigmond et al. (1999, p. 822) note, in the visual system

“high-level neurons classify visual stimuli by integrating information that

is present in the earlier stages of processing, but also by ignoring information

that is independent of that classifi cation.”