Brain damage can occur after stroke (blockage or rupture of a brain artery), tumor (or surgery to remove a tumor), surgery to remove epileptic tissue, viral infection that involves the brain, traumatic head injury (car wreck, fall, gunshot wound, etc), or from diseases like Alzheimer’s or Parkinson’s (whose causes are poorly understood).

Domain-specific deficit: Trouble processing, learning and remembering one particular type of material. For instance, damage to some parts of the left frontal and temporal lobes may cause problems speaking, reading, and comprehending language (aphasia) in addition to problems learning and remembering verbal material. Damage to other parts of the temporal or occipital lobes can leave language intact, but cause problems in understanding the visual world and remembering visual images. But these sorts of domain-specific problems are not usually called amnesia.

Anterograde amnesia: Difficulty learning new material after the brain damage

Retrograde amnesia : Difficulty remembering material that was learned before the brain damage

Most amnesic patients have some degree of both anterograde and retrograde memory loss.

Post-traumatic amnesia (PTA) occurs after a head injury (a fall, a car accident, etc) often gets better over time as swelling and bruising of the brain heal. We’ll focus on amnesic cases that are more permanent – where are the damaged areas, and what is the nature of the memory deficit?

The relevant brain areas for domain-general memory problem

Medial temporal lobe =  hippocampus and surrounding cortical areas. These areas are part of the cerebral cortex, but are anatomically different from the rest of cortex (which is called neocortex).

- Patient H.M. is a famous example of a medial temporal lobe amnesic. He had large parts of the left and right medial temporal lobes surgically removed in the 1950's to relieve uncontrollable seizures.  Read more about H.M. in the chapter by Eichenbaum.

Diencephalon = a cluster of subcortical structures in the center of the brain.  For memory, the most important are the thalamus and the mammilary bodies. 

- Patient N.A. is a well-studied diencephalic amnesic.  He had an accident with a fencing foil that penetrated his brain via his nose.

- Korsakoff's syndromeis a disorder that affects the diencephalon. Caused by long-term alcohol abuse, and the thiamine (Vitamin B-1) deficiency that comes from heavy drinking: the liver is occupied processing alcohol all the time, and doesn't metabolize vitamins (especially thiamine) as well as it should.   The diencephalon regulates activity in the cortex, so many parts of the cortex (especially prefrontal cortex) are under-active in Korsakoff's syndrome.

Classic amnesia involves damage to medial temporal or diencephalic areas.

Basic description of the classic amnesia syndrome: Intellectual abilities are intact, STM memory is intact, LTM is impaired.

Retrograde loss in the classic amnesia syndrome:
Limited in time, goes back for some period before the damage, but not all the way back to birth: time-limited retrograde amnesia. Perhaps there is a long period ofconsolidation as memories grow stronger after the initial learning. The proposal is that many parts of the neocortex learn and store memories, but these areas need the medial temporal lobe and diencephalon to help them lay down permanent memory traces. Once a memory is securely stored in the neocortex, the medial temporal lobe is no longer needed.

1) Learning. Over a period of weeks, animals are presented with 100 different pairs of objects. For each pair, one of the two objects has a food treat under it. Monkeys learn which one of each pair is the "good" one, until they always reach immediately for the correct object in each pair.

2) Variable retention period. For two weeks, or four weeks, etc, up to 16 weeks, the monkeys don’t see the objects again.

3A) Control animals ("N" for normal): no actual brain surgery (they do get anesthesia, as if for an operation, but their brains aren’t touched)

3B) Experimental animals (H+, for hippocampus): have medial temporal lobe removals, kind of like HM’s surgery

4) Test memory for the 100 pairs of objects

- Normal animals (black circles) showed a forgetting curve. 2 to 4 weeks after learning, they did pretty well; 8-16 weeks later they had forgotten some objects, but still performed well above chance (50%)

- Experimental animals had a memory deficit, but ONLY if the medial temporal lobe was removed 2-4 weeks after learning. If surgery happened 16 weeks after learning, they seemed to remember as much as the control animals.

- What these results show is that for about a month after learning, the memories for those objects were dependent on the medial temporal lobe – remove that brain area, and the memory goes too. But after a month, it seems that the memories no longer need the medial temporal lobe – as if they had been "copied" to some other part of the brain.

Anterograde memory in the classic amnesic syndrome

Impaired learning: events, facts, words, faces, visual or auditory stimuli, etc as tested by recall, cued-recall, recognition (direct memory tests)

Spared forms of learning:
1) Habituation: Reduced response to a repeated stimulus. The typical habituation paradigm is to present a startling stimulus – causes a defensive response like flinching, moving away, jumping a bit. But if the stimulus is repeated (and doesn’t cause any harm), this startle response declines.
2) Classical conditioning
3) Perceptual and motor skill learning. A good example of a skill that is both perceptual and motor is mirror tracing: staying on the lines as you trace a drawing you can only see in the mirror.
4) Cognitive skill learning. One laboratory example is the Tower of Hanoi puzzle (see Neath pg 8.6).
5) Motor/somatosensory biases. One example of this comes from biases in judging a weight. Your estimate of how heavy Object X is – from picking it up – will vary systematically depending on what you picked up JUST BEFORE: lifting a heavy weight makes the next one seem lighter, lifting a light weight makes the next one seem heavier.
6) Perceptual priming. Many examples of this in class, and in "multiple memory systems" lectures. See especially indirect tests using word fragments or picture fragments.
7) Artificial grammar learning. Depending on how you do the experiment, artificial grammar learning may be a "cognitive skill", or it may also have a motor component. Here’s an example that is pretty much in the "cognitive skill" category. This experiment was much like the Reber experiments described in Neath: 1) see a bunch of "grammatical" letter strings, then judge new letter strings as either grammatical or ungrammatical. Here, they added a recognition test: present both previously-seen and new grammatical letter strings, and ask for "old/new" recognition judgements. The amnesic subjects were not signficantly different from controls on the grammaticality judgements (so they did learn about the grammar), but were definitely impaired at recognizing actual letter strings they had seen before.

Declarative vs Procedural memory. Note that the list of memory abilities impaired in amnesic subjects corresponds to what is called Declarative memory. The list of learning abilities which are intact in amnesia corresponds to what is called Procedural memory.

This is because the amnesic syndrome was a large part of what led to creating the theoretical dichotomy. See the first "multiple memory systems" lecture for more on the Declarative/Procedural split.

So, what’s wrong exactly in the classic amnesic syndrome? Or, what are these brain areas good for when they work? Three alternative ideas are below, good arguments for all three.

- Rapid learning, learning after one exposure versus repeated practice: This idea is that synaptic change in the neocortex can only take place slowly, but the medial temporal lobe undergoes rapid changes in synaptic strength. A memory is initially stored in the medial temporal lobe, but this brain area helps "train up" the neocortex over time.

- Flexibility of the acquired knowledge. In normal memory, information learned in one way (a printed word for instance) can be generalized (to a spoken word or a picture). This means that the memories are available to more than the original brain area that processed the information (or the printed word would never be recognized when spoken, the memory would be stuck in visual cortex). The widespread connections of medial temporal and diencephalic regions to other brain areas might allow memories stored in one place to be accessed by other brain areas. (The same is true of prefrontal cortex, this part of cortex also has connections to many other brain areas).

- Content of the information to-be-learned: "configural". The question here is "why are so many forms of learning spared in amnesia?" The proposed answer is that those procedural forms of learning can take place in small brain neighborhoods. Motor skill learning only requires motor areas, visual perceptual priming only requires visual cortex, auditory perceptual priming only requires auditory cortex, etc. In contrast, declarative memory involves learning relationships between different sorts of information: learning someone’s name means forming a relationship between what they look like (visual), the auditory sound of their name, and a few facts about them (this person is in my history class, etc). For these sorts of declarative memories, we need to be able to learn and retrieve relationships between bits of information that are stored in different brain areas: this requires the medial temporal lobe and diencephalon because they have inputs and outputs to many other brain areas.

Damage to prefrontal cortex also causes domain-general memory deficits, but these are more subtle than real amnesia:

1) Recall moderately impaired, but recognition is fine

2) Source memory impaired. Source memory is roughly "memory for the context in which something was learned", remembering HOW you learned something. We may be able to talk more about this later in the semester.

3) Some frontal patients (but not all) confabulate – make up stories when they can’t actually recall, and seem to believe their own stories.

Figuring out exactly how the frontal lobe contributes to memory is an active research area. But a broad idea is that it is critical for controlling and monitoring memory, that you need prefrontal cortext in order to mount an effective search of memory contents (so that recall is impaired after frontal damage), and that you need it to evaluate the products of a memory search (is that right? have I remembered everything relevant? should I check on this? etc)