Stephen van Vlack
Sookmyung Women`s University
Graduate School of TESOL
Human Learning and Cognition
Spring 2006
Week 8 Answers - Terry, Chapter 7 & Lamb, Chapters 12 and 13
1. What is Dual-Store Theory and how does it explain memory?
Dual-Store Theory is one of several models used to explain memory. It proposes a system of memory that encodes information in either short-term memory (STM) or long-term memory (LTM). In this model, information processing actually begins with what is called the sensory memory, which contains data received by any of the five senses. Information in sensory memory is retained for a very brief time unless it is recoded into short-term memory. STM is also brief, lasting only 15 to 30 seconds under lab testing conditions. It has limited capacity, codes items in verbal form, and is subject to verbal rehearsals. Moreover, STM is displaced as soon as a new item is introduced for processing, which explains why forgetting occurs easily in this store. With maintained rehearsals, however, STM can be encoded into LTM. On the other hand, long-term memory is relatively permanent and virtually limitless in capacity. It is more durable and can be encoded in modalities other than verbal, such as images, smells, touch, and emotion. Because LTM stores vast kinds of information, theorists hypothesized further subdivisions of LTM.
This leads to subsequent dichotomies, such as episodic vs. semantic and explicit vs. implicit memory. Episodic memory is a personal memory system that consists of autobiographical accounts that are marked by specific temporal and contextual information. Semantic memory is a store for general knowledge, which is also referred to as dictionary or encyclopedic knowledge. This includes facts, words, language, and grammar. Episodic and semantic memories are sometimes categorized under explicit memory because knowledge in this system can be recalled, declared verbally or explicitly. Implicit memory, on the other hand, is performance that occurs independently of any conscious attempts to recall. This type of memory is not accessible to conscious verbal recall and is acquired through several means that include procedural learning and priming. Procedural learning is knowing how to do things through the acquisition of generalized rules. It involves perceptual, motor, and cognitive skills, all of which contribute to allow a person to perform a task without necessarily knowing verbally the sequence.
Priming is facilitating access to information through previous exposures to the same data. This concept upholds the notion that we remember better those things that we have been previously exposed to or have been "primed" to recall. The concepts of STM and LTM in Dual-Store Theory have found evidence from different laboratory experiments and clinical studies, thus reinforcing the heuristic value of this approach to explaining human memory.
2. What variables affect the stages and the process of memory and how can they be used for better memory and learning ?
Memory plays a pivotal role in how the human brain works in relation to learning. From how to get to your home after class or which subway station you need to get off at, to what answer should be given in exam or how to deal with the strangers you meet on the way, all of them are connected to leaning based on memory processes. Then, why can`t we remember what we experienced or learned at the very moment we need it? This question (number 2) is about how to facilitate memory and remove deficits of memory, that is forgetting.
Terry deals with the reason why we cannot recall something, have some problems in memory and what the reason of forgetting may be by analysing the steps and process of memory. First, one possible reason for forgetting can be found in the different stages of memory. Information that we get everyday goes through to be formed (encoding), retained in the memory system of the brain (storage) and later retrieved (retrieval).
The most important aspect in encoding is rehearsal which represents recall or use the of memory, just do it repeatedly. Provided more rehearsal with slower pace and immediate test, a memory can be strongly formed. Once well formed, forgetting can be found in retrieval and problems in retrieval can exist in both encoding and storage. If retrieval is facilitated by providing reminder (reinforcement) cues which act as prompts to aid recall, encoding deficits can be cured. Intention according to exposure the participant to enough information. It means memory should repeatedly be exposed to prior conditions. Discrimination of a recent memory from those of earlier lists can be one of the reason for retrieval deficit. Encoding or retrieval deficits tell us about the importance of presentation or input and how to deal with the activities in the classroom. If the previous exposure of the target language with rehearsal is provided enough, memory or learning can be enhanced and forgetting can be delayed.
The other reason of forgetting can be found in processing approaches of memory.
In cognitive processing, forgetting is caused by shallow processing. In contrast to shallow processing, sustained retention means deeper or more elaborate processing. According to the book, processing is co-related to rehearsal such as maintenance rehearsal and elaborative rehearsal. Maintenance rehearsal means the passive repetition of information that is repeating something over and over, on the other hand, elaborative rehearsal represents a deeper level of processing and more active form of processing which involves meaningful analysis, comprehension of the material by using mnemonic devices, forming mental images and relating the to-be-recalled material to existing knowledge. Elaborative rehearsal lead to longer retention.
Focusing on process of the memory, method of teaching or learning should be varied for enhancing memory and longer retention by using meaningful materials and interaction.
In retrieval, we try to find out the memory of environmental conditions that presents at encoding. This is transfer-appropriate processing stated reinstatement of the cognitive operations that were used at encoding. According to the experiments, implicit and explicit test of memory are not separate system but processing is different. Implicit memory depends on a match between perceptual operations at encoding and retrieval. It is said that test of memory should follow the way of the presentation.
3. What are the basic principles of connectionist models and how do they combine to explain the process of learning and forgetting ?
Connectionist models are based on several basic principles. First, as these models were founded on the principles of neuroscience and the physical brain, there is a great reliance on the neuron and the connectivity of the brain. Furthermore, these neurons combine to form a neural network through which electronic and chemical messages can move with varying degrees of speed and efficiency. The speed and efficiency of the connections is determined by the degree of use the connection undergoes. These neurons and the neural networks they form have a resting level and activation level determined by the degree of electronic and chemical signals required to fire them. Priming is an important factor in the firing of neurons and neural networks. As electronic and chemical signals activate intended neurons, surrounding neurons (with proximity determined by pathways, not physical distance) receive signals that combine to lower the remaining amount of electricity required for activation. These models go a long way in explaining learning and forgetting. As we have stressed throughout this whole semester, connections and associations are the keys to learning. This idea is well supported by connectionist models. Additionally, forgetting can also be explained as the connections either become inefficient due to a lack of use or simply never gained the minimally required strength at the initial building stage (due to a lack of reinforcing information).
Here are several key components of Connectionist Models and key reference words related to each point-
1. Each neuronal unit can potentially have connections to many other units. *Neural networks, Connectivity, Neural Pathways*
2. The strength of connections increases with pairings of active neural units and weakens
without the activation of others. *Delta Rule*.
3. The neuronal unit or network can be activated. *Activation, Threshold Levels, Chemical and Electronic Signals*
4. Activation may require stimulation from multiple input units. *Threshold Level*.
5. Multiple layers. *Hidden layers*
4. How is the cognitive system creative? (L12)
Creativity is an interesting thing, but it does not necessarily work in the same way that analytical linguists make us think about it. Creativity is not the simple ability to be able to make new or novel utterances, which is the view of Chomsky and other analytical linguists. For them the ability to slot new words into the framework set up by the computational system (UG) is creative. Not really. That is simply slotting lexical items into pre-made positions in a structure. Another way of looking at creativity which analytical linguists have acknowledged is the ability to create new words or the process of word coining. This is indeed a highly creative process but it is also something which occurs rarely in relation to overall language use.
The cognitive linguistic view takes a more comprehensive view of creativity. In this view, and remember this is based on the belief that language and thought are linked, creativity of language and thought involve the same processes. Obviously, as with everything else in the brain it is all about connections. In this view, the cognitive system is creative in that it allows new combinations to be made between networks already in brain but previously unconnected. New patterns on different levels are created, learned, and used and it is easy to see these cognitive connections through language. An example might help.
(1) William could feel the beer eating away at his thoughts.
This is not a particularly striking sentence. People say things like this all the time without ever really thinking about how they do it or what they are doing. And we also have no difficulty interpreting this as readers or listeners either. It is an example of creativity of the type cognitive linguists posit. We have a metaphorical extension of the word `eat`. In the reality of the world beer is not able to eat anything. Beer is inanimate and incapable of conscious action. Thoughts, fo their part, are an abstract entity (if they are even an entity of any kind, which is highly debatable) and as such cannot be eaten simply because they have no form. So what we have here is an example of creativity in that we have taken aspects of certain concepts and connected them together in order to create a new meaning. Mostly, certain aspects of the verb eat are being connected with both beer and thoughts. None of these things are connected in the real world, yet we create meanings by doing so in the brain. This creativity, as is all creativity, is achieved by connecting things. It should be noted that only some of the many possible features of eat are being carried over and applied to beer and thoughts. This should be clearer in the next example.
(2) Did you see the hunchback of Notre Dame skulking over there in the corner.
This may be a costume party, but then again maybe it isn`t. In this case the person being talked about is identified as the hunchback of Notre Dame as through this identification/designation has come to take on certain features of that fictional character. They might not even need a hunchback, but may be just be hunched over a bit in addition to wearing certain dorky or old fashioned (maybe dirty) clothes. Not all the features of the hunchback of Notre Dame are going to be put on this unfortunate person, but just some. This is what Fauconnier (1997) calls partial mapping. The example is also compelling because we can see how lexis is affected by or maybe even generates the mapping with the use of the verb skulking. Well, we assume that the hunchback of Notre Dame skulks. The use of this lexical item shows how entire lexical schema can be involved in the mapping and in this case the mapping might have been triggered by the use of the word skulk. This is how people are creative. They link things together which have not been previously linked. And this is the type of creativity which is not only used in the creation of language.
As we mentioned in class we explain problems with computers through an analogy to the viruses that biological entities have. In reality a computer virus is quite unlike the cold you might catch, but we label them in the same way because they is the only mechanism that humans have for explaining things: we take the things we know and are familiar with and then reapply them in new circumstances. This same type of thought can be seen in things as seemingly disparate as paganism and science. In both of these models of explanation people look at the evidence around them and based on what they know and are familiar with come up with the most sound explanation possible. They do not create new things. They simply reassign new tasks to old workers. If a new type of development is seen in the physical world the pagan will carefully consider and with the precision of a scientist determine which of her/his gods is responsible. Scientists will do the same based on their own set of scientific knowledge. And once the decision has been made for explaining this new phenomenon then a vast amount of other knowledge is brought to bear in the scenario. In the pagan scenario a specific and very important tree falls down, it must have been struck by lightening. Thor is the god of lightening. He must be angry at us or at another god, the god of trees Frigg. Based on what we know about Thor and Frigg and their past then we can determine how to act or how the story will unfold. In a science scenario we are all aware that problems in our computers are often the result of viruses. Based on what we know about computers and viruses like the flu then we can determine how we will act to save the computer. In this way, we can see how these extensions these examples of creativity really do affect our thoughts and behavior.
As this relates to language and language learning the thing we need to bear in mind as teachers is that language is flexible. It is not at all fixed. The constraints on language change and language use (it should be obvious at this point that they are the same thing) are societal in general and not strictly linguistic. The ability of the speech community you are in to understand your novel utterances ids the main constraint and this is not based on syntax at all but on semantics. Language change is driven by meaning and syntax (as we shall see in questions 6 and 7) is a much less important aspect of that. As language teachers we need to focus on meaning first and lexis as the main purveyor of that meaning as well as link to other linguistic aspects needs and deserves the greatest focus. Language learners have been shown a tremendous disservice by having their focus put on syntax and not on meaning (lexis) because the former is taught as being rigid while the latter can never be so.
5. How are new connections created? (L12)
According to Lamb (1999), there are basically three ways that new connections are created.
New connections are made by linking previously unlinked neural nets.
This is the type of thing that we generally think about when we think of forming new connections. Two neural networks which were previously unconnected are now connected in some ways. We saw some examples of how this happens in the examples of creativity above. It is important to remember that this occurs as a result of being creative; in seeing (both literally [with the ocular system] and proverbially [through concepts] something new through the rather simple process of feature hopping. All of these processes of creation discussed above are like learning (for they are a type of learning) physical. They are reliant upon physical changes in the brain. In this case dormant or latent connections are made active (are recruited) to make a link between two previously unconnected neural nets. Some of these will be more permanent and other less so. If your friend John, for example, breaks his leg and this causes him to walk with a pronounced limp then it may be possible to call him gimpy. In this case new connections are made between John and the lexeme gimpy. If John was never gimpy before then the link simply did not exist previously. As soon as John heels and no longer walk with a pronounced limp then it is not longer funny or appropriate to call him gimpy. In this case the connection will slowly degenerate because it is not used. So, only some of these new associations made will survive. Some will degenerate because the situation which fostered their development disappears or becomes less important (note: this may or may not be an example of habituation) or they might degenerate because they are rejected by others as implausible. If the response to using this connection is slightly negative, then it will dissipate (note if the response is strongly negative then the link will be encoded strongly and if not used much will survive as a possible alternative).Most of these new links will simply not make it long term, but some will and that is how language changes.
New connections are created by strengthening previous links.
The idea here is that there were already links present between two different neural nets, but the links were weak and not often used. As new information is brought in to the equation the older, weaker links can be rejuvenated. We all have a vast amount of available memories in our heads and these are physically represented by very weak to inactive links. As these links are reactivated by new experiences which incorporate elements of some past experiences these old pathways can be strengthened and make the information in the networks accessible. In this way we can see that the difference between available and accessible memories is the strength of the connections in the associated networks. It is not how the information was learned (Krashen`s learned vs. acquired information), but how it is encoded and stored. It should be noted that any neural network is going to be composed of thousands of not millions of neurons. The connections between different neural nets will never be entire, this means that there will only be connections or string connections between certain parts of the associated networks. In the examples above where only certain information is connected between any two neural networks we can see how this works in the reality of language use.
New connections are created by making changes in the threshold functions.
The previous way of storing new creative uses (learning) basically focusses on the connections themselves, that is the route which carries the information and we understand at this point that a stronger, better connection allows information to be carried faster and for that reason, better connections are the preferred routes, thus making them even stronger in a never-ending cycle. Here, now, we are not dealing with the connections themselves but the threshold levels of both the neurons and the synapses with the neurons has strewn along its axon, up to 50,000. Each neuron has a threshold level, which is, as stated in question 3 above, the amount of energy needed to get the neuron to fire. When a neuron fires it releases its energy as well as chemicals which either enhance or inhibit the flow of the energy over synapses. Just as each neuron has its own threshold level, which rises and falls depending on use, each synapse also has something similar related to the chemicals present in the gap. The constant presence of neuro-enhancers like dopamine (as a result of frequent passage or firing of the parent neuron) means that the synapse is constantly ready to transport the electrical impulse set out from the neuron. They same applies to inhibitors. This simple equation determines more or less which of the many different synapses a neuron has will be used to transport information and to which neural nets. Changing these threshold levels and functions will change which synapses are being used more and ultimately will determine where the information from the neuron is being sent.
6. What are some different types of linguistic patterning and how are they formed? (L13)
There many different level/types of linguistic patterning we can think of. We can list them as:
Phonological/morphemic/lexical/cognitive
Each of these patterns is formed through previous use. With enough use, this use which is perceived and stored, becomes usage. This is the crux of what is termed a usage-based model of language and language learning. Usage-based models are basically the developmental guise of cognitive linguistics. In essence, language usage, as most simply represented by patterning like the SOV pattern in English or the SOV pattern in Korean, comes as the result of language use. As with everything else this is a result of connections. A neural net is always associated with a larger neural net incorporating more and larger bits of information. This the key. It may be easy to understand how this simple approach applies to individual words (a neural net involving articulatory gestures is associated to other such networks in a particular order), but the same simple system applies to things above the word level as well. We can create sentences and novel sentences because we already have whole sentences stored in the brain (as prototypes possibly, see Fillmore, 2003 chapter 4). It is these `sample` sentences which guide our production and reception. They are the basis of larger patterning.
7. What are some of the conflicting elements of patterns and how do they survive long-term? (L13)?
In the brain there are two dynamic forces which through their competition are able to determine much of what neurons do. They are the elemental forces of activation and inhibition and without these two there would be no such thing as human intelligence or any intelligence at all. As was mentioned above, Neurons are complex entities with connections not only running out (though the axon) but also running in (through the dendrites). It might help to think of neurons as dams. They are able to facilitate or inhibit the flow of information (like a dam) and not just generally but (unlike any dam) to specific places. Neurons are sorters and it is dual ability of neurons as both inhibitors and enhancers which makes this possible.
A PDP model can be likened to a race where several racers start out. As they get to checkpoints (certain neurons) the quickest racers will be allowed to continue on in the race (and they will even split into new racers) while the slower ones will be inhibited from moving forward (remember that inhibiting a specific synapse will leave it inoperable for quite some time). In this way only strong connections used a lot previously will be allowed to be used again. In this way well-established patterns are strengthened while less frequent ones are extinguished through the process of inhibition.
Looking at these dual forces, the only real forces available to neurons we can also explain bilingual uses of language and why only one language is generally available for use at a time. Once a particular language is selected for use or is encountered in use, then other neural networks, undoubtably connected, will be inhibited. In a balanced bilingual such inhibition is relatively easy because they have networks of different but equal strengths in both languages, but for other types of bilinguals using certain languages, especially languages that are normally inhibited, will be very difficult not because the information is not there but because it is somewhere inhibited.