Imagery in the Brain

Psy315h - Class 25

(This page last updated 25 November, 2005.)
  

After years of debate on the nature of mental image representations, research moved into the area of Cognitive Neuroscience.

Review Positron Emission Tomography (PET) discussed earlier in the semester.

PET study to determine whether mental imagery activates brain areas also used in vision.

Kosslyn, Alpert, Thompson, Maljkovic, Weise, Chabris, Hamilton, Rauch, & Buonanno (1993).

• Like many imaging studies, it used the subtractive method to try to determine which brain areas were responsible for particular cognitive functions. 
• A PET scan produces a measure of the amount of activity for many different regions across the brain. To make these measures interpretable, they are organized into "slices" which can be displayed as images. To produce each image, the different activity levels are mapped to different colors.
• Each PET scan can take a number of minutes, so PET experiments must use cognitive tasks that can be repeated over a number of minutes. Usually, each subject performs one task during one scan, and another task during another scan.
• The subtractive method has its limitations. It requires that two tasks differ only in the presence of one cognitive component, which is almost never true.
• Kosslyn and colleagues are trying to prove once and for all that visual images use analog representations. Recall that V1 and other visual areas early in visual processing are organized retinotopically. This pattern suggests that they are spatially organized, or analog representations.
• If those areas are active while subjects make visual images, it would strongly suggest that analog representations are used in imagery.

   

  Perception Task Subject sees letter on grid. In one square of grid is an X. Subject decides whether X is in a square covered by the letter and responds by hitting button.
  Imagery Task Subject sees empty grid with X. Also sees lower case script letter. Must image upper case block letter in grid and decide if X is in a square covered by letter and respond by hitting button.
  Sensory-Motor Control Subject sees grid with X. Hits response button. This task has almost the same sensory input and motor output as the imagery task, but no image.

• One group of subjects performs the perception and imagery tasks.
  • By subtracting perception from imagery, they can see what brain areas are only active during imagery.
• Another group performs perception and sensory-motor control.
  • They will perform a similar subtraction with these subjects to show that any effects must be due to imagery.
• Visual imagery apparently activates an area in the occipital lobe. This may be part of V1, but it is near the border and could be one of the areas next to V1.
  • Most brain regions here are retinotopically mapped (at least in monkeys), so it is probably a retinotopically mapped area.

Kosslyn et. al. tried another experiment to show more convincingly that imagery activated retinotopically-mapped brain structures.

• Each subject imagined letters as small as possible for one scan, and as large as possible for another.
• The letters were presented auditorially, and subjects had to decide whether or not each letter had any curved lines.
• A relevant fact about area V1:
  • Area near the fovea is represented in the most posterior part of V1. Peripheral areas of the visual field are represented in the anterior part of V1.
• When subjects make small images, more activity in posterior V1 (the foveal area). When subjects make larger images, more activity in anterior V1 (the peripheral area).
  • Note this activity is only in the right hemisphere.

next class: Availability and Representativeness
 
Psych 315H: Cognitive Psychology Kyle Cave Psychology Dept. U. Mass.