Short-term and Working Memory

Definition of memory The processes involved in retaining, retrieving, and using information about stimuli, images, events, ideas and skills after the original information is no longer present. Important implications of this definition: Memory includes learning Memory involves a variety of processes that can function with autonomy

Importance of memory Obviously being able to remember past experiences and learned skills is important for ability to make decisions, etc. in the present Memory also important in predicting the future. Much of what we know about the future results from our knowledge of the past

Stage theory of memory External  sensory  short-term  long-term stimuli memory memory memory Attention and sensory memory covered in the previous section Now we turn to the study of short-term memory (STM) and then later to the more encompassing working memory

Limited capacity ofSTM Miller (1956) proposed the magic number 7 + 2 We can only receive, process, and retrieve approximately 7 pieces of information at a time His study asked people to recall in order lists of numbers of varying length

Overcoming this limitation Chunking – organizing or grouping individual pieces of information into a single “chunk” 18005551212 1-800-555-1212 Today referred to as recoding 1,8,0,0 is recoded as 1-800. 4 pieces of information is recoded into 2 pieces

Recoding Recoding can occur in STM if there is the time and mental resources available to reorganize the information Using long-term memory to recode information – mnemonic devices Using a well learned strategy to recode information An example is verbally recoding information because language usage is over learned

How accurate is this magic number 7+ It is accurate for relatively simple information groups digits, words, etc. Not as accurate for more complex information Example: with 3 or 4 word phrases the magic number becomes 3 to 5

Decay from STM Brown-Peterson task: Subjects shown three letters and then a 3 digit number Subjects told to count backwards from the number given by 3’s until asked to recall the letters Counting backwards prevented the rehearsal of the letters Results: 3 second delay – little over 50% retained 9 second delay – little over 20% retained 18 second delay – less than 10% retained

Interpretation of Brown-Peterson Memory loss in STM is the result of decay; “the memory trace decays” without rehearsal STM different than long-term because it was believed that forgetting in long-term memory results from interference

Can intereference occur in STM? Could the counting backwards have actually interfered with memory – not just preventing rehearsal Reexamination of Brown and Peterson data (Keppel and Underwood (1962)) Waugh and Normal (1965) – Was the memory loss the result of the passage of time- more loss as more time passed Or was increasing the amount of counting backwards interfering with retention?

Keppel and Underwood (1962) They saw that on the first trial, memory performance was nearly perfect As subjects participated in more trials their performance declined Their conclusion: previous trials interfered with later trials – proactive interference

Release from proactive interference Changing the nature of the items to be remembered reverses the decline in performance due to proactive interference Wickens et al, 1963 Two groups of subjects given 3 trials following the Brown-Peterson task (letters) - Memory performance declined with each trial Control group given a 4th trial using letters Experimental group switched to remembering digits Experimental group, but not control group, performed perfectly; they were released from proactive interference

Waugh and Norman (1965) Subjects verbally presented with lists of 16 digits – some lists were presented at a rate of 1 digit per second others at 4 digits per second The last digit was the repeat of an earlier digit. Subjects asked to write down the digit that followed the earlier digit. 4, 2, 6,8, 9, 2 correct answer is 6

Waugh and Norman (1965) One group of lists took 16 seconds to present the other group took 4 seconds If decay causes loss of information from short-term memory, the 16 second group should remember less because more time would have passed before they responded Problem for decay theory was there was no difference between groups. With no interference performance was the same

New decay theory Interference theory appears to fit the data better than decay theory Active decay in a special situation – subjects switch from one task to another and must “forget” the previous instructions

Altmann and Gray (2002) Subjects shown one number at a time 1, 2, 3, 4 or 6, 7, 8, 9. 1 group of trials asked is the number odd or even? 2nd group of trials asked is the number from the group of large numbers or small numbers? One group of subjects were switched very frequently, the other infrequently Frequently switched group had faster reaction times and were more accurate Conclusion: forgetting previous decision rule was faster in this group because they needed to remember the new rule – old info actively decayed

Recall and the serial position effect Present subjects with a list of 20, 30 or 40 items 1 every second, and ask them to recall them in order. Primacy effect – more of the 1st items presented are remembered Recency effect – more of the final items are remembered 1st items rehearsed long enough to get in long-term memory; last items still in STM

Recall and the serial position effect II Glanzer and Cunitz (1966) Same study except subjects told to count backwards after list given Recency effect disappeared, not primacy Glanzer (1972) Again same study except subjects given 3, 6, or 9 seconds between each item – longer to rehearse Increase in primacy effect, no increase in recency

Purpose of STM Rehearsal important part of STM Rehearsal maintains a memory trace for a short period of time Rehearsal helps transfer information from STM to LTM

Retrieving information from STM Donder’s reaction time studies – 1880’s Subtractive tasks A - a simple reflex – see light-push button B – decision + reflex – see blue light- push button – see red light don’t push button C – decision + choice + reflex – see blue light push button 1 – see red light - push button 2 How long does it take to make a decision? Subtract time to perform A from time to perform B

Sternberg Task Problem with Donder Subtractive tasks – there could be an interaction between A and B such that the reflex might not be the same with the decision as when it is alone Sternberg invented an additive task

Sternberg Task (cont.) Subjects shown a list of letters ranging from 1 letter to 6 letters, then shown a single letter as a memory probe. They were to respond as quickly as possible indicating if the letter was in the list or not Reaction time was recorded Two important variables were involved The number of letters in each list The location of the letter in the memory probe – in the beginning, middle, or end

Sternberg Task (cont.) Three possible results: STM is searched in a parallel fashion – if true then length of list or location should have no effect STM searched in a serial fashion, we search until we find the letter – both length and position important All of STM is searched and then we make a decision, a serial exhaustive search – length would have an effect location in the list would have no effect

Results of Sternberg Task 1 letter list - 37.9 ms 2 letter list – 75.8 ms Each additional letter increased reaction time by 37.9 ms Location of the letter in the list or if the letter not in the list had no effect Conclusions: we scan all of STM before making a decision Many limitations found to this research, but it led to major advances in cognitive sciences

Coding information in STM Baddeley (1966) – information coded acoustically or verbally Subjects asked to remember either a 5 word list or a 10 word list Remembering 5 word list STM; 10 word list exceeds STM and is LTM In all lists, the words either sounded alike (cat, hat, cat) had similar meanings (tiny, small, little) or were unrelated

Baddeley (1966) Results: 5 word list errors – most errors were made when words sounded alike – house instead of mouse. Fewer errors on lists with similar meaning or unrelated 10 word list most errors with semantically similar words – labor instead of work Conclusion: Similar sounding words confused in STM because memory code was acoustic. Semantically similar words confused in LTM because memory code was using meaning

Wickens (1972) – Release from proactive interference Proactive interference occurring as a result of semantic coding in STM 5 groups of subjects given 3 trials of lists of 3 words each all from the same category Group 1 – names of fruit Group 2 – vegetable names Group 3 – flower names Group 4 - names of meats Group 5 – names of different professions Then all groups given a 4 trial where all list contained names of fruit

Wickens (1972) – Release from proactive interference Results: 1st trial all groups about 90% correct 2nd trial all groups about 50% 3rd trial all groups 35 – 45 % 4th trial professions 80%, meat 50%, flowers 47%, vegetables 40% and fruit 32% Conclusion: Information was coded using semantic information causing groups to confuse current list with previous lists

Visual coding in STM Mental rotation task of Shepard and colleagues Subjects shown 2 objects and asked if they were the same or different in different orientations Interpretation people held the 1st figure in STM and mentally rotated the 2nd to make a comparison Objects were either different or the same but rotated to a different orientation Subjects took longer to answer when the object had been rotated further 600, 900, 1200

Working memory Developed as a result of STM memory not being useful in explaining how short term memory processes were used in problem solving Also finding that some people with brain damage can have impaired STM – a digit span of only 2 items, but no deficits in learning, comprehension, or memory

Components of working memory Executive control system – planning, initiating, and integrating information – high cognitive abilities Two subordinate systems: Articulatory or phonological loop – rehearses verbal information – auditory and semantic coding Visual-spatial sketchpad – maintains images and spatial representations – visual coding

The working memory process Central executive gives subordinate systems information to hold until it needs it again Example in textbook: (4 + 5)2 3+ (12/4) Central executive does (4+5) 2 = 18 sends answer to articulatory loop to remember while it calculates 3+(12/4) = 6 It then retrieves 18 to calculate 18/6 = 3

Limited capacity of working memory The subordinate systems have few attentional resources; when they are involved in a demanding task they must get resources from central executive These resources are limited Creates the concept of dual tasks – two tasks being performed at the same time The validity of working memory can be tested using dual task methods

Dual task studies Subjects given reasoning tasks of varying complexity – performed by the central executive Then asked to perform different secondary tasks that were similar in articulatory demands but whose memory requirements differed It was found that the most complex reasoning tasks were most effected by secondary tasks that required the most memory resources As the articulatory loop took more resources from the central executive, it found solving the complex reasoning tasks more difficult

Other supporting data for working memory 8-arm maze Neuropsychological evidence Damage to areas of the left frontal lobe creates deficits in verbal working memory Damage to areas of the left frontal lobe creates deficits in spatial and visual working memory PET scans have shown that the Dorsolateral pre-frontal cortex is most active when working memory task are performed, same left and right distinction