Monday, August 30, 2010

Peepholes Into a Coherent World: BrainWorks Series

To most of us, the brain is a "black box." It creates a seemingly coherent world with which we can interact, but we have only limited understanding how it does what it does. But when focal brain damage happens to persons, neurologists and neuroscientists can begin to see how the brain puts the pieces of the puzzle together to create the world.
It's here that people with visual agnosias come in handy. Behrmann had previously studied people with integrative agnosia, who have difficulty recognising and naming complex objects as a whole, and instead seem to pay unusual attention to their individual features. One person, for example, mistook a picture of a harmonica for a computer keyboard, presumably thinking the row of air-holes in the mouthpiece were computer keys (Journal of Experimental Psychology: Human Perception and Performance, vol 29, p 19). Others have mistaken a picture of an octopus for a spider, and a pretzel for a snake.

In 2006, Behrmann put one of her patients, known as SM, through a series of experiments alongside people with normal vision. All were shown a set of three-dimensional objects on a screen, each made from two simple geometric shapes. Afterwards, the volunteers were shown a stream of these images, with a few new objects thrown in. Their task was to report whether or not they had seen the objects before.

While those with normal vision performed with nearly 100 per cent accuracy, SM made some intriguing mistakes. He knew he hadn't seen an object before if it contained a new part, but those that had the same parts in a different configuration confused him. About half the time he mistook these for the familiar objects (Journal of Experimental Psychology: Human Perception and Performance, vol 32, p 1169).

To Behrmann, the results suggest that our brains normally construct objects from a series of smaller building blocks, which she calls our "visual vocabulary". To recall our concept of an object, she says, we form a mental map of the way these parts fit together. It was at this stage that SM failed. "He had a good representation of the parts, but understood little of how they were combined," Behrmann says. _NewScientist

Advanced brain imaging is helping neuroscientists to sort between the different the different varieties of visual agnosias. This helps us understand the different functional brain modules, and to learn where they are located.
Brain scans have revealed that people with visual form agnosia tend to have damage to the ventral (lower) part of the brain's visual area. People with optic ataxia, on the other hand, have damage to the dorsal (upper) part. This led to the idea that we have two streams of visual processing. The ventral pathway is necessary for perceiving or recognising an object, while the dorsal pathway deals with an object's physical location in our visual field and, if we need to perform an action on it, guides the movement of our bodies. For this reason, scientists often refer to the two processes as the perception-action, or the what-where, streams of visual processing.

...In fact, the closer neuroscientists look, the more modular our visual systems appear. MRI scans of people with and without agnosias have suggested that within the ventral stream, separate aspects of appearance are processed independently. This year, psychologist Cristiana Cavina-Pratesi at Durham University in the UK found that shape, texture and colour are all processed in individual regions (Cerebral Cortex, DOI: 10.1093/cercor/bhp298).

Yet our experience feels markedly different. When we consciously see something, all these disparate elements are stitched seamlessly together, so we know instantly that an apple is smooth, green and round. The question of how we accomplish this is central to the study of conscious perception.

...So important is the role vision plays in most people's everyday lives that most research has concentrated on visual agnosias. Now the hunt is on for similar disorders that affect the other senses. Recently, for example, neurologists found a person who could understand speech but not other sounds. Coslett, meanwhile, is investigating whether simultanagnosics also have trouble binding other sensory sensations together, such as sights and sounds.

Now you see it...

There are many visual disorders, typically caused by damage to specific parts of the brain.
  • Simultanagnosia - Seeing only one object at a time, even when viewing a scene comprising many items
  • Integrative agnosia - Inability to recognise whole objects, tending to focus instead on individual features of an object
  • Visual form agnosia - Inability to describe the shape, size or orientation of objects, yet exhibiting no problem in manipulating them
  • Optic ataxia - Ability to report the shape and size of an object, though attempts to manipulate it are clumsy
  • Prosopagnosia - Failure to recognise the faces of familiar people
  • Pure alexia (aka agnosia for words) - Inability to identify individual characters or read text, even though subjects are sometimes able to write
  • Agnosia for scenes - Inability to recognise known landmarks or scenes
  • Colour agnosia - Ability to perceive colours without being able to identify, name or group them according to similarity
_NewScientist
We have discussed the binding problem before, but it is important to begin to zero in on the parts of the brain which are involved in binding different aspects of reality together into a "coherent whole."

It is also important to begin to learn the actual dynamic mechanisms which are responsible for creating the actual "sensation of consciousness." Or, "The Feeling of What Happens," as Antonio Damasio expresses it.

It is only by delving deeply into these processes that we will be able to conceptualise ways in which we may profitably reverse-engineer a human brain. But that will mean letting go of the "algorithmic theory of conscious intelligence" which has waylaid so many well-meaning artificial intelligence researchers in the past.

Scientific American has a short piece on a parallel topic

No comments:

Post a Comment

LinkWithin