3.5 Iconic Memory

Having laid out these four levels of visual processing, we can now discuss

visual cognition as being dependent upon iconic memory, attention, and

the synchronous fi ring of the neurons in the areas relevant to the visual

percept. Iconic memory is implicated in visual cognition because the visual

representation has to be held in mind, so to speak—if even for a very

short time—so that unifi cation of the various parallel processes that contribute

to cognition can occur. The idea is that if the disparate pieces of

information were too fl eeting, then there would not be enough salience

in the visual fi eld to be able to encode the incoming information. In

studies where individuals are shown quick fl ashes of objects on a screen

and then asked to identify them, if the object does not remain on the

screen for at least 250 milliseconds (a quarter of a second), then the

individual does not “see” the object or thinks that the test had not begun.

This is so because the object is fl ashed on the screen at such an incredibly

fast rate that the visual system is not even aware of the occurrence (Barrett,

Dunbar, & Lycett, 2001; Julesz, 1983). Through PET scans, activity in the

left prefrontal cortex has been shown to be active in individuals who

attempt to encode visual information (Kosslyn et al., 1999a, 1999b,

2001).

Further, the existence of iconic memory acts as a bridge between noncognitive

visual processing and cognitive visual processing. We are wholly

unaware of the processes in the lowest level of the visual hierarchy comprising

the trajectory from retina, through LGN, to V1. Iconic memory is

kind of like the paused scene on a movie you rent or a snapshot photograph

in your mind that enables you to be aware of or cognitive of a visual

scene—although such a scene or photo can be held in mind for only an

incredibly short amount of time.

When we think of memory, what comes to mind are recollections of

mental maps, events, sights, sounds, smells, and the like. Further, these

recollections are not a jumbled mass of confusion but are organized into

coherent picture-like scenes. With respect to visual memory, these scenes

are, in effect, visual images that we have stored somehow in our brains

and deliberately can recollect if necessary. So how does this storage take

place?

Following leads from Hebb (1949, 1966), Bliss & Lømo (1973) demonstrated

that the hippocampus of the mammalian brain exhibited alterations

in synaptic strength of neurons, depending upon the amount of

stimuli applied. They termed this alteration in synaptic strength long-term potentiation (LTP). LTP could last for hours in brain slices, while in the

intact animal it could last for days. Malenka & Siegelbaum (2001, p. 419)

note that LTP is a “ubiquitous property,” since it has been observed in the

excitatory synapses of a variety of areas of the brain, including the hippocampus,

all layers and areas of the cortex (including the visual cortex), the

amygdala, the thalamus, and the cerebellum.

In addition, Kandel’s (1976) groundbreaking work associated with the

gill-withdraw refl ex of the marine snail, Aplysia, showed that simple forms

of learning—habituation, sensitization, and classical conditioning—could

produce changes in the synaptic strength of neurons resulting in memory

storage. It seems clear that there exists some kind of storage mechanism

at the cellular level. This is something important to keep in mind, as the

organs and processes of the brain are comprised of these collections of

neurons (cf. Maviel, Durkin, Menzaghi, & Bontempi, 2004; Fink, 2003;

Hoffman & McNaughton, 2002).

However, this is not the full story concerning memory storage. Studies

of patients whose association areas are damaged have shown that different

representations of an object are stored differently. For example, a

person suffering from associative visual agnosia whose posterior parietal

cortex has been damaged can identify objects by drawing them but

cannot name them. Conversely, a person suffering from apperceptive visual

agnosia whose occipital lobes are damaged can name objects but cannot

identify them to draw them (Farah, 1990). Another example is prosopagnosia,

the inability to recognize familiar faces or learn new faces that

results from damage to the IT cortex. People who suffer from prosopagnosia,

although unable to process or recall faces, still can process and

recall other objects, such as animals and tools (Geschwind, 1979). These

studies indicate that the visual image is a product of multiple representations

in the brain, each having their own neural correlates and each

concerned with a different aspect of the visual image. This implies that

there is no one deposit memory storage area. Rather, there are multiple

storage areas, and recollection is itself a process of building up disparate

pieces of information.

In this section, I discussed a storage mechanism in the visual system

for at least two reasons. First, such a mechanism is necessary in information

exchange so that a receiving afferent entity actually can be infl uenced

by the information communicated by an efferent entity. Think of

an action potential and neurotransmitter release in the synaptic cleft of

a neuron A and how that affects a neighboring neuron B, possibly causing

another action potential in neuron B. Or, think of the DNA transfer when neurons differentiate. In both of these processes, the effects in

some way must be imprinted within the efferent entity so that the communicated

information evokes some kind of change in the efferent entity.

These processes are representative of the myriad exchanges of information

taking place throughout the entire nervous system.

Second, a memory mechanism is integral for visual imaging. When I

want to deliberately recall my spouse’s face, a process begins whereby the

IT cortex, hippocampus, and other temporal cortical areas are activated.

That I can recall my spouse’s face means I had to imprint her face, and

this process also involved the IT cortex, hippocampus, and other temporal

cortical areas (see Kosslyn, 1987; Kosslyn et al., 2001). Alternatively, say I

wanted to hold an object in my hand—like a shirt off the rack at a department

store—and analyze an aspect of it to check and see if it has any

imperfections. In order to accomplish this task, some kind of temporary

memory would be necessary to sustain its image long enough in my visual

fi eld.

Having laid out these four levels of visual processing, we can now discuss

visual cognition as being dependent upon iconic memory, attention, and

the synchronous fi ring of the neurons in the areas relevant to the visual

percept. Iconic memory is implicated in visual cognition because the visual

representation has to be held in mind, so to speak—if even for a very

short time—so that unifi cation of the various parallel processes that contribute

to cognition can occur. The idea is that if the disparate pieces of

information were too fl eeting, then there would not be enough salience

in the visual fi eld to be able to encode the incoming information. In

studies where individuals are shown quick fl ashes of objects on a screen

and then asked to identify them, if the object does not remain on the

screen for at least 250 milliseconds (a quarter of a second), then the

individual does not “see” the object or thinks that the test had not begun.

This is so because the object is fl ashed on the screen at such an incredibly

fast rate that the visual system is not even aware of the occurrence (Barrett,

Dunbar, & Lycett, 2001; Julesz, 1983). Through PET scans, activity in the

left prefrontal cortex has been shown to be active in individuals who

attempt to encode visual information (Kosslyn et al., 1999a, 1999b,

2001).

Further, the existence of iconic memory acts as a bridge between noncognitive

visual processing and cognitive visual processing. We are wholly

unaware of the processes in the lowest level of the visual hierarchy comprising

the trajectory from retina, through LGN, to V1. Iconic memory is

kind of like the paused scene on a movie you rent or a snapshot photograph

in your mind that enables you to be aware of or cognitive of a visual

scene—although such a scene or photo can be held in mind for only an

incredibly short amount of time.

When we think of memory, what comes to mind are recollections of

mental maps, events, sights, sounds, smells, and the like. Further, these

recollections are not a jumbled mass of confusion but are organized into

coherent picture-like scenes. With respect to visual memory, these scenes

are, in effect, visual images that we have stored somehow in our brains

and deliberately can recollect if necessary. So how does this storage take

place?

Following leads from Hebb (1949, 1966), Bliss & Lømo (1973) demonstrated

that the hippocampus of the mammalian brain exhibited alterations

in synaptic strength of neurons, depending upon the amount of

stimuli applied. They termed this alteration in synaptic strength long-term potentiation (LTP). LTP could last for hours in brain slices, while in the

intact animal it could last for days. Malenka & Siegelbaum (2001, p. 419)

note that LTP is a “ubiquitous property,” since it has been observed in the

excitatory synapses of a variety of areas of the brain, including the hippocampus,

all layers and areas of the cortex (including the visual cortex), the

amygdala, the thalamus, and the cerebellum.

In addition, Kandel’s (1976) groundbreaking work associated with the

gill-withdraw refl ex of the marine snail, Aplysia, showed that simple forms

of learning—habituation, sensitization, and classical conditioning—could

produce changes in the synaptic strength of neurons resulting in memory

storage. It seems clear that there exists some kind of storage mechanism

at the cellular level. This is something important to keep in mind, as the

organs and processes of the brain are comprised of these collections of

neurons (cf. Maviel, Durkin, Menzaghi, & Bontempi, 2004; Fink, 2003;

Hoffman & McNaughton, 2002).

However, this is not the full story concerning memory storage. Studies

of patients whose association areas are damaged have shown that different

representations of an object are stored differently. For example, a

person suffering from associative visual agnosia whose posterior parietal

cortex has been damaged can identify objects by drawing them but

cannot name them. Conversely, a person suffering from apperceptive visual

agnosia whose occipital lobes are damaged can name objects but cannot

identify them to draw them (Farah, 1990). Another example is prosopagnosia,

the inability to recognize familiar faces or learn new faces that

results from damage to the IT cortex. People who suffer from prosopagnosia,

although unable to process or recall faces, still can process and

recall other objects, such as animals and tools (Geschwind, 1979). These

studies indicate that the visual image is a product of multiple representations

in the brain, each having their own neural correlates and each

concerned with a different aspect of the visual image. This implies that

there is no one deposit memory storage area. Rather, there are multiple

storage areas, and recollection is itself a process of building up disparate

pieces of information.

In this section, I discussed a storage mechanism in the visual system

for at least two reasons. First, such a mechanism is necessary in information

exchange so that a receiving afferent entity actually can be infl uenced

by the information communicated by an efferent entity. Think of

an action potential and neurotransmitter release in the synaptic cleft of

a neuron A and how that affects a neighboring neuron B, possibly causing

another action potential in neuron B. Or, think of the DNA transfer when neurons differentiate. In both of these processes, the effects in

some way must be imprinted within the efferent entity so that the communicated

information evokes some kind of change in the efferent entity.

These processes are representative of the myriad exchanges of information

taking place throughout the entire nervous system.

Second, a memory mechanism is integral for visual imaging. When I

want to deliberately recall my spouse’s face, a process begins whereby the

IT cortex, hippocampus, and other temporal cortical areas are activated.

That I can recall my spouse’s face means I had to imprint her face, and

this process also involved the IT cortex, hippocampus, and other temporal

cortical areas (see Kosslyn, 1987; Kosslyn et al., 2001). Alternatively, say I

wanted to hold an object in my hand—like a shirt off the rack at a department

store—and analyze an aspect of it to check and see if it has any

imperfections. In order to accomplish this task, some kind of temporary

memory would be necessary to sustain its image long enough in my visual

fi eld.