Stephen van Vlack

Sookmyung Women`s University

Graduate School of TESOL

Human Learning and Cognition

Spring 2006

Week 11 Answers - Terry, Chapter 10 & Lamb, Chapter 17 & Lieberman (2000) Chapter 1

1. How long is the long-term memory and what is the nature of storage? (T10)

One of the important things we need to remember about long-term memory is that not everything we perceive makes it into long-term memory. Imagine how difficult life would be if it did - talk about clutter!! Not all memory is good and not all forgetting is bad. Then, what is forgetting?

Forgetting is construed as either the actual loss of information from storage or a loss of access to information that is still in storage somewhere. Of these two options both are probably true but there has been more support for the latter according to more contemporary views. Then, what causes forgetting to occur? Both the time of decay and interference play a part in forgetting. In other words, memory weakens and fades overtime and proactive interference and retroactive interference also affect memory loss. Proactive interference refers to effects from memories preceding the target memory and retroactive interference indicates influence from memories formed after the target memory was encoded. There is counter-evidence on the time effect. For example, according to Jenkins & Dallenbach (1924), sleeping enhances memory, but this is probably due to a stabilization and even rehearsal effect present during sleep in addition to the simple observation that sleep prevents any possible interference.

Duration of memory is also important issue here. Ebbinghaus.

According to psychological models of semantic memory, items are organized hierarchically and linked together in a top-down, or bottom-up fashion. This is a familiar proposal from Lamb (1999) so we should have no difficulty understanding this. The basic assumption is that general knowledge in semantic memory is common among individuals sharing the knowledge and personal memory is idiosyncratic. Based on this, two models are have been proposed; the traditional hierarchical model and the alternative model. Another point of view (and this is often given as an alternative, but we take it to be complementary to the hierarchical model) is the neuro-psychological disassociation model. This model claims the also familiar proposal that changes in the activity at the synapses between neurons affect memory storage and retrieval. For us, because we have been looking more closely at this from Lamb (1999) and other cognitive linguistic concerns, it seems that both proposals are true and necessary to explain the neurological basis of both the brain and language.

Much more is learned than is typically recalled in long-term memory. At this point, we can distinguish between the available memories that are in storage and the accessible memories that we can actually recall. According to Tulving and Pearlstone (1966), some of the "forgetting during the initial attempt to recall" was due to memories that existed but could not be retrieved. The idea of retrieval failure suggests an interpretation of forgetting that differs from a theory that memories spontaneously decay and disappear from storage.

2. What are some of the variables that affect retrieval and how can we enhance retrieval? (T19)

The three main factors that affect retrieval are: the distinctiveness of the memory, practice at retrieving the memory, and the presence of effective retrieval cues.

Events that are distinctive, flashbulb memories are well recalled and in the isolation effect, an item that differs from the remaining items in a list is better recalled. Why are distinctive events better retrieved? One explanation is that their retrieval cues uniquely target a single memory. By contrast, a retrieval cue might be too broad, potentially retrieving many items.

Retrieval is facilitated by previous retrieval. Taking a test shortly after studying is better retrieved than to test retention some time later like studying the week later. This is called testing effects. The additional study primarily benefits information that is not recalled after 1st studying. The effects of prior testing are greater if the same test is given both times. Semb and colleagues (1993) had students take an end of course exam after completing a psychology course, one for 4 months later and the other 11 months later, the 11 months later test score was 10 percent higher and more difficult prior testing is also more effective. Glover (1989) found an initial free-recall test produced better retrieval later than did an initial cued recall test, which in turn was better than an initial recognition test. More recall on the retest given later is more effective. Hypermnesia means remembering that actually improves over successive attempts at reproduction of the studied material and this is the opposite of the forgetting that we expect to occur over time or amnesia. Erdely and Kleinbard (1978) tested 60 object line drawings for study 20 times over the following week. Whereas just over 26 items were recalled on the first test, 38 were recalled on the final tests. Bahrick and Hall (1993) reported hypermnesia occurred for other test materials including recall of general information, foreign-language vocabulary, and names. From one test to another, some items are indeed forgotten but the number of recovered items on successive test exceeded the number lost, leaving hypermnesia as a net gain. Why would people remember more and not less across successive tests? First additional test provide more time to retrieve and second during free recall they prompt their memories with subjective or self generated cues. People think of one thing or another to help themselves remember because everything is interconnected in the brain. Self-prompting cues vary across successive tests. Different things come to mind at different times so the later cues tap items that were not adequately cued on the earlier tests. The capacity for additional recall of material across a series of test has implications for eyewitness testimony. The assumption that the first recall is the most valid and anything remembered later is suspect is not right because additional accurate information can be elicited through repeated recall attempts.

There are two kinds of cues: Cues that have strong preexisting associations to the target memory and Cues that were encoded along with the to be recalled item when it entered into memory

Encoding specificity: Cues encoded along with to be recalled item when it entered into memory. The best retrieval cues are those that were also present and encoded with the target. Recall is facilitated when encoding and retrieval condition are matched. Matching cues between encoding and retrieval are called Tulving'""""""""""""""'Contextual learning: Remembering will be better if the cues present at encoding are also available at retrieval. Contextual cues are sometimes overshadowed by more explicit retrieval cues and interaction of the context with the to be remembered items may be required. Contextual cues of time and place, for example, English learning only in the classroom and English use out of class and taking the same seat in the class.

Mood-dependent recall: Stimuli arise from moods and emotional states, and that these mood stimuli can enter into associations. Mood-specific recall can have implications for depression and other affective disorders. Being depressed may lead to the retrieval of mostly sad memories, which only perpetuates the depressive mood.

Although memory is inherently variable, some ways of testing it are more sensitive than others. The most common example of this situation is the different performance people display on recognition versus recall tests. For instance, students almost always claim that multiple-choice questions are easier than fill-in-the blank questions.

To sum up, recognition is better than recall. Although this phenomenon could be attributed to the greater number of cues a recognition memory test usually provides, there are complications. For instance, subjects can use mnemonic strategies to generate additional cues and so improve their performance in free recall. Just how well a subject does on a recognition test depends on the context cues in which the test is given and the difficulty of the distractors. Thus, the exact level of performance in recall and recognition tests can depend on many factors.

Interactions between Study and Test

Implicit Measures of Memory

3. What are metamemory and retrieval failure? What are their characteristics and implications in learning? (T10)

Metamemory and retrieval failures

The Definition of Metamemory

Metamemory refers to people's knowledge, awareness, and control of their memory. Simply put it, metamemory means knowing what you know, knowing how your memory works, and being able to assess your own memory. Metamemory is a developmental occurrence: as we get older, we get better at remembering, and we also improve our abilities for developing strategies to assist our memories (Bruning, et. al., 1999). We often discover that certain strategies must be developed to improve memory. For example, list making, mnemonics, post-it notes, memo pads, sensory associations, and many other strategies serve to establish cues that will assist in the processing of larger chunks of information, thereby making retrieval easier. Metamemory skills provide the student with awareness of strategies for recalling information as well as the ability to use their repertoire of strategies. It is assumed that a sophisticated learner knows a number of strategies or processing rules, such as the following:

a. Single Item Repetition: Repeating material over and over, one item at a time.

b. Cumulative Rehearsal: Repeating material over and over in a cumulative fashion, rehearsing old items along with new ones (e.g., Flavell, 1970).

c. Meaningful Organization: Looking for meaningful, semantic relationships among items. (e.g., Moely, 1977).

d. Hierarchical Allocation: Studying information in order of its importance, with more important information studied first (e.g., Brown & Smiley, 1978).

e. Differential Effort Allocations: Expending more effort to study the material that is not yet learned (e.g., Brown & Smiley, 1978).

f. Imagery Elaboration: Making up interactive images that include the to-be-learned items (e.g., Pressley, 1977).

g. Verbal Elaboration: Making up a story to include the to-be-remembered items (e.g., Rohwer, 1973).

h. Keyword Method: Transforming unfamiliar items (e.g., foreign words) to more familiar ones (e.g., sound-alike English words) and then putting them into relational images with other information (e.g., Pressley, Levin, & Delaney, 1982).

i. SQ3R: Surveying what is to be learned; questioning oneself; reading the material; reciting it; reviewing all important information (e.g., Robinson, 1961).

Retrieval failures

Sometimes we fail to retrieve information that we know we have. Retrieval failure is the idea that people can have information in memory that they cannot retrieve at the present time. Retrieval failure is explained as the result of a difference between memories being available and accessible. All the experiences that you have ever had may have been encoded in your memory, and if they were, they would be part of your available memories. At any time, however, you can only retrieve a small part of those memories, and that small part is the memories that you have accessible. The way to prevent retrieval failure is to increase the accessibility of a particular memory. One way to increase the accessibility of a particular memory is to duplicate the context in which the memory was encoded. The typical retrieval failures are feeling of knowing (FOK) and Tip-of-the tongue (TOT) phenomena. TOT means the state in which information is available but not accessible from memory. In TOT, usually parts of the information are accessible, but not enough to warrant a response. Retrieval blocking and incomplete activation are some reasonable explanation of TOT states. Retrieval blocking explains that activation of items in memory that are similar to the target compete with the target during a memory search. Thus, the retrieval of the target is suppressed and related words serve to block retrieval. Incomplete activation says an initial memory cue may not activate a target word or name enough for retrieval of target and related words facilitate eventual retrieval TOT states can be resolved by using various search strategies, rather than sticking to a single strategy.

False memory (retrieval)

A false memory is a memory that is a distortion of an actual experience, or a confabulation of an imagined one. Many false memories involve confusing or mixing fragments of memory events, some of which may have happened at different times but which are remembered as occurring together. The cause of false memory: The more associates there are in the list, the more likely false retrieval becomes.

4. What are some of the different variables related to connections and how are connections created? (L17)

The basic idea here is that after describing his purely theoretical (top-down) model which he calls Relational Network Representation, Lamb (1999) decides that he is going to put his money where his mouth is and compare it to what researchers have found out about functional neural systems in the brain. This is an interesting endeavor in trying to fit things together and associate the conclusions Lamb and others have made about language simply by looking at language use. It is important to realize that this entire book (Lamb, 1999) is a highly abstract theoretical endeavor to describe language in a way linked to but not actually affected by the reality of the brain. Now, in the last couple of chapters Lamb has turned to the actual neuroanatomy of the brain (the neocortex exclusively) to see how well his model stands up to scrutiny from reality. Now we are moving bottom-up. Hopefully now we understand better why the Lamb book is so hard to understand at times. In effect it builds a model of language based on the brain without making direct reference to the brain and often without including very many examples form language.

So what we have in this chapter is a description of the brain in relation to the model we have already read about. Lamb goes to great pains to try to convince us that his model does indeed match the structure and functioning of the brain even though it was developed separately. For us, much of this is not so important. It is nice to hear that these things really do exist and find out what they are really like, but at this point we already have bought into certain aspects of the model. A particularly compelling example of how this works plays out in the distinction between local and distributed representations, which in the end is seen as being necessary and an important way of coding information in the brain, for we do indeed need category headings.

5. How might conceptual categories be represented by a neural network? (L17)

I think we already have a well-developed idea of how a conceptual category would be represented by a neural network. Basically, a conceptual category like CAT is represented in the brain through a multitude of interconnected neural networks. Each of these networks carries a bit of information which is part of the concept. These different networks are connected directly to a local distributed network via long distance connections. The job of the local network is to pool all the different connections coming in so that information coming in from below can be either fanned out to different information necessary for understanding the intended meaning or, for information coming down from above, it can be funnelled together so it can move effectively to the applicable neural networks below which enable articulation. It should be obvious from this description that while the system is reliant on the spreading activation model because this model allows lots of rather disparate information from around the cortex to be associated, there would be to a certain extent a necessary hierarchical organization of the different types of neural networks involved.

6. What role does the basal ganglia play in the brain? (Li1)

The basal ganglia may be the key to human language and cognition, because they connect to the cortex and thalamus and organize muscle-driven motor movements of the body. In other words, the basal gangliaconnect to cortical areas to participate in motor control, the regulation of affect, language, and cognition. The basal ganglia control voluntary movement with the cerebellum. Cerebral cortex sends information to the basal ganglia, and then the basal ganglia send information right back to the cerebral cortex via thalamus. Cerebral cortex exchanges information with the cerebellum in the same way. However, the output of basal ganglia is different from that of the cerebellum. The output of basal ganglia is inhibitory, while the output of cerebellum is excitatory. The balance between two systems enables us to controlprecise movements like picking up the grains of rice with chopsticks. Thus, the function of the basal ganglia is often described in terms of "brake hypothesis."For example, in order to sit still, you must put the brakes on all movements except the reflexes that maintain an upright posture. To move, you must release the brake on voluntary movement.

7. How can learning be explained through a distributed neural network?(Li1)

We may say that learning is making output based on the association of new information with existing, relatively old informationdistributed on the neural network. Herein, learning involved structural change and modification of synapticweights of neurons. The distributed neural network models are redundant. Information like a representation of one object, for example, an apple, is distributed here and there in the neural network. The information may also exist in the neural network that represents another object sharing common properties like an orange or a red light. Therefore, the representation does not disappear right after a certain network loses its function or dies. What happens in the neural network is structural change and connection weights. According to Hebb(1949), what we have to consider is the connection weights of neurons, not just individual working of neurons. According to Rumelhart(1986), the pathways that transmit signals more often attain higher conduction values. In other words, when we learn new things, connection weights of neurons and the conduction values of pathways change.

8. How does the functional neural system relate to motor control and vision? (Li1)

Neural systems in the brain are functional. Lieberman says that the functional neural system is a network of neural circuits that work together to perform a certain behavior or task. And also the functional neural system evolved to produce a rapid motoric response to enhance survival. Some evidences for functional neural systems can be found in studies that monitor electrophysiologic activity in the brains of monkeys. These experiments show that a class of functional systems exist that rapidly integrate sensory, cognitive, and motor activity to achieve particular motoric responses to external stimuli. According to the Neurophysiologists who study vision using the brains of monkeys, no single and unitary map exists in the primate brain to cope with a particular activity. Instead, different functional systems exist each of which achieves particular goals. The experiment of Charles Gross and his colleagues(1995) show that one system appears to be adapted to grasping objects that are moving toward a monkey's face. Their studies of the macaque monkey brain show that cells in the putamen, a subcortical basal ganglia structure, respond vigorously when a monkey sees small objects approaching its face and eyes. About 25 percent of these sites also respond to tactile sensations on the monkey's face. These putamenal neurons, in turn, communicate to neurons in the monkey's ventral premotor neocortex. Although specialized areas of the cortex process incoming visual information, no general-purpose visual module exists that processes all visual input and then sends appropriate vectors to a motor interception module.

9. What are the roles of the primary motor cortex in relation to learning? (Li1)

The primary motor area of the cortex is the major control region for planning and initiating voluntary movements. Scientists who studied the primary motor cortex of monkeys found that it is also organized in a functional manner. But many texts still represent the organization of the human primary motor cortex by showing a picture of an upside-down body(toes, feet, hand, fingers, lips, tongue, etc) in which different areas of the motor cortex control different parts of the body. However, the organization of the primary motor cortex is not somatopically modular, although general regions of the primary motor cortex control head, arm, and leg movements. On the contrary, It has been claimed that individual muscles are influenced by neurons in several separate locations in the motor cortex. Moreover, individual cortical neurons have branches linking them to other cortical neurons that control multiple muscles. It is likely that neural elements that make up the functional processing units continually get reorganized while producing an adaptive behavior. In fact the primary motor cortex appears to be adapted for learning and storing motor control programs of a new task. Repeated trials appear to shape neuronal circuits. The process is apparent when a baby learns to walk. At first walking requires total concentration and is slow and inaccurate. After repeated trials, walking becomes automatic and rapid. Similar effects occur as we learn to drive, catch balls, or talk.