Topic 5: Short-Term Memory/Working Memory/Attention Flashcards
What is short-term memory?
much of the work done on what we might think of as “temporary” memory led to the development of the modal memory model
conceived of short-term memory (STM) as simply being a passive buffer that allows for the temporary retention of information that hasn’t been committed to long-term memory
What is working memory?
was proposed as a system that can temporarily hold onto information, as well as manipulate/process that information (in a way that wasn’t account for in the STM model)
What is the support for the model modal of memory?
prolonged neural activation related to mechanisms involved with direct perception (e.g., activity in visual cortex)
the activation of semantic representations
regions/networks involve with cognitive control
What is STM in the digit span task?
STM has conventionally been measured using a digit span test
Miller’s landmark publication argued 7 +/- 2 to be the “magic number” (corresponding to the average digit span)
How did H.M. perform on the digit span task?
was able to keep things active i STM indefinitely via rehearsal (among other things, demonstrating that “short” in this context doesn’t refer to the amount of time that has passed since something was encoded)
What is the distinction between STM and WM?
WM emphasizes the active manipulation of information being temporarily held in memory (vs. passive recall, e.g., digit span)
depending on how we define/conceive of the two concepts, we could consider tests of WM to necessarily include a STM component (though the opposite is not true: tests of STM don’t necessarily require the active manipulation of information), which could introduce various confounds
it’s also the case that some historical studies which used the term STM may have bee actually/also measuring WM
What is the Jacobsen (1936) lesion study?
bilateral surgical removal of the frontal cortex in monkeys resulted i chance level performance on a delayed-response task
they concluded the prefrontal cortex is responsible for “immediate memory”
methodological note: monkeys demonstrated intact visual/motor/motivational abilities (when no delay was imposed in the same task), ruling out a few potential confounds
How did Malmo (1942) replicate the results of Jacobsen (1936)?
Malmo (1942) replicated the Jacobsen result, while also extending the results by introducing a new condition: a lights off condition
surprisingly, the monkeys performance dramatically increased in the lights off condition (~85% accuracy), while remaining close to chance levels for the lights on condition (~50%)
why might this be? this hits at contributions of the PFC related to cognitive control
What is the Ghent et al. (1962) study on PFC damage?
found no obvious impairments in a sample of humans with PFC damage during a digit span task, yet did find detects in the same task within a sample with lesions in the left temporoparietal cortex
What is the Milner et al. (1962) study on PFC damage?
found performance impairments in a human sample with PFC lesions, though only while using repeating (“closed”) stimuli sets (as compared non-repeating (“open”) stimuli sets)
difficulties with repeating (“closed”) stimuli sets can cause problems related to proactive interference, e.g., if you know you’ve seen an item during the experimental session but can’t remember when (was it this block or the last block, which could change what the correct response is, etc.)
What is an explanation for why PFC damage sometimes disrupts STM but not always?
one possibility is that, rather than being critical for the simple temporary retention of information, the PFC might be important for the control of STM, and/or the control of behavior guided by STM
in this view, the PFC may not necessarily be the location where information is temporarily retained (the “buffer”), yet may nevertheless contribute to the ability to temporary hold onto information elsewhere in the brain
What was the Kubota and Niki (1971) study?
took recording from neurons in the monkey PFC for delayed response
found two “classes” of task-related activity
those that became active during the delay (presumed to be responsible for “holding” the info in memory, i.e. acting as the buffer)
those that became active just prior to their response period (presumed to relate to motor responses)
What was the Fuster and Alexander (1971) study?
recordings from neurons in the monkey PFC (and mediodorsal nucleus of the thalamus: MD) for delayed recognition
neurons in both PFC and MD fired continuously throughout the delay period
What was the Rushworth et al. study?
found that STM for the retention of simple object representations was not impaired by bilateral lesions to the inferior convexity of the PFC
that result seems inconsistent with the proposition, based on the earlier electrophysiological study we just discussed, that sustained activity in the PFC may be necessary for the temporary retention of information/representations
What were the findings of the Petrides et al. study on frontal lobe sulcus lesions?
lesions did not impair performance for a task involving the simple retention of basic object representations across long intervals (90 and 120 seconds)
the result seems consistent with the Rushworth et al. study, implying the PFC may not be absolutely required for all STM-based tasks
however, the lesions did impair performance for a delayed non-match to sample task across short intervals (10 seconds), which requires more complex processing (remember the square during a delay, choose the thing that isn’t a square during the test, e.g., the circle)
this result seems consistent with the studies we discussed in which impairments were found when the lights were on (but not off), and with closed (but not open) stimuli sets, implying the PFC may be important for certain kinds of STM-based tasks requiring more complex processing
What are the discrepancies for the role of the PFC in STM and WM?
the results of the studies seem to converge on the idea of a role for frontal areas in controlling STM, though those areas may not necessarily be the location of the neurons that actually hold onto the information, i.e., the buffer (or at least not always)
Petrides et al. also found that anterior IT lesions had the opposite effect, representing a double dissociation (impaired the simple retention of basic object representations but not performance in the delayed non-match to sample task)
could that region be the location where some representations are temporarily held onto, with the PFC being the location that controls those neurons?
How are the discrepancies for the role of the PFC in STM and WM reconciled?
one possibility is that the PFC may not be the only region that can support “an explicit neural representation for the sample stimulus”
for example, some studies have observed sustained activity in IT cortex and entorhinal cortex, which may be important in certain situations
continued activity in sensory networks may serve a similar purpose
as a potentially relevant parallel here, consider our previous discussions about how activity in sensory and/or motor regions may contribute to LTM representations
What are the what/where pathways?
Goldman-Rakic suggested the what/where pathways (ventral/dorsal streams) may make differential contributions for holding different kinds of information in working memory (information about identity vs. space)
working memory may be organized along similar functional lines as the visual system itself
perhaps this could explain the implications for a role of the temporal lobe in the temporary retention of information
What is the role of sustained firing in STM and WM?
one possibility is that sustained activity might not always be important for STM/WM processing
another possibility is that the high threshold for achieving statistically significant activation in a typical univariate fMRI analysis (i.e., non-MVPA style analysis, just looking at “hot spots”, or regions of interest; ROI) may obscure some contributions from relatively weak signals
What is the sensory recruitment hypothesis?
proposes that STM/WM may (at least in part) be supported by the same systems that handle basic perceptual processing
What was the Serences et al. (2009) the sensory recruitment hypothesis?
used a 10 second delayed recognition for gratings (lines of a particular orientation), superimposed on a colored background
plot at left shows most responsive voxels in V1 (primary visual cortex), and we see the BOLD signal drops to baseline during delay (i.e., no apparent sustained activity)
on the surface, that would seem to suggest there is no sustained activity occurring in V1 during the delay interval…
however, while there was no apparent sustained signal associated with V1 found in these analyses, the researchers were able to successfully train a pattern classifier to discriminate between some basic stimuli types
these results reinforce the fact that, just because no “statistically significant” sustained activity is observed (e.g., in the BOLD response), doesn’t necessarily mean that there is none occurring
What is the Emrich et al. (2013) study on the role for sustained firing?
applied MVPA while testing STM for direction of motion, using a field of red, blue, or green dots
load was manipulated by varying how many colors of dots were moving
rostral intraparietal sulcus (IPS) and various parts of PFC were found to be “load sensitive”, meaning that changes in the load manipulation were correlated with changes in activation
Based on the Emrich et al. (2013) study, does finding that something is load sensitive always implicate STM/WM?
their MVPA analysis, combined with attempts to train a pattern classifier, found that the areas with the greatest sustained activation didn’t actually carry much/any stimuli-relevant information
in this case, larger loads may have required more effort, created more stress, etc. which may have accounted for the “load sensitive” activity
What is a change detection paradigm?
this involves simply asking participants to attend memory array, then test them on whether they can correctly identify the test array as the same or different (similar to the idea of a digit span test)
typically use 50/50 same/old stimuli for test array
can manipulate working memory load (more squares), stimuli properties to retain, etc.
What is the Vogel et al. (2005) study on using ERPs to study WM?
assessed participants’ WM capacity and split them into two groups them based on that factor (high vs. low capacity)
during the main experiment, participants were shown either simple or cortex stimuli (with or without distractors) while ERP responses recorded
high-capacity participants demonstrated less of a response to the distractors
suggests the high-capacity participants more efficiently ignored (filtered/inhibited, etc.) the distractors
could this group difference be attributable to superior executive control that is often associated with larger working memory capacities?
What kind of ERP results would load theory predict?
load theory predicts that distractors tend to get automatically processed, provided enough residual cognitive resources are available (whereas distractors will not be processed when resources are “maxed out”)
Can we use the contralateral delay activity (CDA) ERP as an index of working memory load?
the ERP seems to increase in amplitude as more is held in WM
in other words, as with the Emrich et al. (2013) study just discussed, demonstrates load sensitivity
What is neglect?
damage to one hemisphere (frequently a stroke in the right hemisphere) can result in unilateral neglect
involves a lack of awareness of information on the side contralateral to the damage
sometimes involves the temporoparietal junction (TPJ), frontal eye fields (FEF), or the caudate nucleus
What is the Chatterjee (2003) study on imaging neglect?
reported a patient who had stroke-related damage to their posterior inferior parietal lobule and posterior superior temporal gyrus
What was the Bisiach and Luzzatti (1978) study on how neglect affect memory?
asked patients with neglect to imagine standing in a highly familiar location (in Milan, where they have spent their entire lives)
counted the number of landmarks they described while imagining facing one direction, then did the same when they turned around
found the neglect symptoms extend to memory and/or imagery, demonstrating that it affects more than just direct visual perception
What is the premotor theory of attention?
proposes that what we think of as attention can be understood simply as an extension of the motor system (an relates to planning, e.g. preparing to make an eye movement, reach for something, etc.)
What is the effect of attention on the AEP (auditory-evoked potential)?
can observe an effect of attention with the AEP (auditory-evoked potential)
attending to a series of beeps increases the amplitude of the N1 and P1, relative to reading a book and not attending to the beeps (i.e., ignoring them)
we also see similar boosts with the visual-evoked potential (VEP) and somatosensory-evoked potential (SEP)
What accounts for this apparent “amplification” effect on attention evident in ERP studies?
more neurons responded while attending
the same neurons responded while attending, yet fired more vigorously (i.e., increase their firing rate)
given that louder sounds produce higher firing rates of neurons, we may be able to think of the effect of attention as being similar conceptually to “boosting” the energy/intensity of an environmental stimulus (which seems to fit with an intuitive understanding of attention: i.e., “turning up” the volume)
What is overt and covert attention?
overt attention refers to attention that is coupled to eye-movements (e.g. shifts in attention related to eye movements), while covert attention involves shifts in attention that are “decoupled” (unrelated) to eye-movements
What is the Di Russo et al. (2003) study on the neurophysiology of sensory attention?
asked participants to fixate centrally ad count the umber of flashed checkerboard stimuli that appear in (only) one of four locations
attention allocated to particular (quadrants in) place boosts at VEP’s associated with stimuli processed in those locations
notice that the high temporal resolution of ERP allows us to say something about exactly when effects of attention “kick in” (somewhere around just prior to 100 ms after stimulus presentation?)
What is a multiplicative increase in a neural signal as a function of attention?
many ERP studies find what could be considered evidence for a multiplicative increase in a neural signal as a function of attention
this fits with the idea that attention merely “boosts” a pre-existing signal
What is the listening to music analogy regarding a multiplicative increase in neural signals?
if you increase the volume while listening to a song, it boosts the signal when there is anything but silence, yet does not increase the volume during silence… this is because it does not “add” anything to the signal, it merely multiplies what is already there
What were the methods of fMRI studies on the neurophysiology of sensory attention?
the experimental design shown on the left involves alternating between periods of time in which nothing is on the screen and when stimuli are being displayed (either presented sequentially or simultaneously, SEQ or SIM)
participants are told to prepare for the stimuli’s appearance by attending to a particular region of space during attended blocks (ATT), while not being given these instructions during unattended blocks (UNATT)
What were the results of fMRI studies on the neurophysiology of sensory attention?
in contrast to the idea of a multiplicative signal, some fMRI studies produce results that can be interpreted as indicating that attention can produce additive increase in signals
evidence that is consistent with his proposition include what are referred to as baseline shifts, in which simply attending (eve before any stimuli was presented) in and of itself produced an increase in the BOLD signal
What is a confound in the results of fMRI studies on the neurophysiology of sensory attention?
imagine that we ignore the (at least intended) mental state that should be present during the interval of time in which participants are supposed to be attending to a particular spatial location (because they are waiting for a target to appear)
if that is used as a baseline because no stimuli is being presented at that time and used as a basis of comparison for intervals of time when stimuli are on the screen (i.e., subtractive logic), could that introduce a confound?
