Download Remembering What Matters

University Studies 15A:
Consciousness I
Memory
Planning for the Last Day of Class: A Concluding Debate
It seems like I really ought not be talking at you all the time, so I would like
to use the last day of class as an opportunity to set up a multi-sided
debate about the usefulness of neuroscience in explaining consciousness.
I believe there is a wide range of positions one might take on the matter,
and of course for the final paper I will ask you to set out your particular
position in relation to visual consciousness, so that you can use the
arguments you develop in your paper.
I would like you to discuss with your classmates, identify other people
who hold similar positions, and form a group to present your arguments
To make this easier (?), I have activated the message board for the class:
https://eee.uci.edu/toolbox/messageboard/m11926/
Memory
As Baars and Gage make clear, memory is not a single entity. There are a
number of important subtypes made possible through different brain
systems:
1. Working memory: usually functions over a period of seconds
2. Episodic memory: first-person memory of singular events
3. Semantic memory: memory about things and events generalized from
episodic memory
4. Procedural memory: memory about how to do things
Does anyone recall whose picture appeared at the beginning of the lecture?
If I had flashed it at a faster rate and immediately covered it (masked it) with
another picture, you still would have what is called an “implicit memory”
that could be tested by asking you to guess the picture.
There are other forms of implicit memory that mostly focus on patterns
extracted from more explicit forms of memory but about which one has no
conscious awareness.
Knowledge of the grammar of one’s native language is perhaps the best
known example.
I will not worry about the distinctions between implicit and explicit memory
during my presentation.
Working Memory
If you read Baars and Gage carefully, you may have noticed a rather different
tone to their discussion of memory and a shift into different level of analysis.
They become far more tentative: the account of working memory is very
much a work in progress. For example:
The abstract for the paper, published this year, states:
Working memory refers to keeping track of ongoing mental processes and
temporary memory. One hypothesis is that this form of memory consists of
multiple domain-specific components…. [Later] advances have often been
interpreted as supporting the alternative hypothesis that working memory
consists of a single, limited-capacity domain-general system for control of
attention…. I argue that the multiple-component perspective and the singleattentional-system perspective are complementary, with each best suited to
asking different research questions, and that many areas of contemporary
debate regarding the nature of working memory reflect differences that are
more apparent than real.
1. Working memory refers to keeping track of ongoing mental processes and
temporary memory.
2. One hypothesis is that this form of memory consists of a variety of subsystems
that work on different modalities: vision, sound, movement, etc.
3. Another hypothesis is that working memory is a form of executive function.
People consider Working Memory (WM) to be a collection of various subunits
because one can disrupt visual WM with a visual task (following a dot moving
on the screen) and verbal WM with a verbal task (repeating the word “the”).
However, one cannot disrupt verbal WM with a visual task nor visual WM with
a verbal task.
These are observations at the level of psychophysics, which (to quote
Wikipedia) “quantitatively investigates the relationship between physical
stimuli and the sensations and perceptions they effect.”
When one shifts to the neuroscience, several questions immediately arise:
1. How does the brain keep that memory (i.e., the pattern of firing of neurons)
intact for the required period?
2. What, for that matter, is that memory? What neurons are firing where?
3. How do you store that visual image of Einstein, for example? We know
something about the visual system: one cannot simply copy an “Einstein
neuron.” Can we copy and store both the image and the knowledge “That’s
Einstein?”
And of course there are the lesion studies which show that if one damages the
tempero-parietal region, working memory is severely impaired. Why?
Baars and Gage present what is in fact a synthesis of the most recent work on
WM as a brain system. They follow Ranganath in proposing:
That is, WM requires focused attention to objects and to relations between objects.
These are two separate functions:
One keeps activating an object (represented in neural networks for highlevel representational spaces like IT)
The other keeps together a multi-modal pattern of activation that includes
multiple networks
These two functions are the roles of the dorsal and ventral PFC
This attentional component seems to be the bottle-neck in working memory:
only four separate “objects” can be maintained at a time by the ventral
prefrontal cortex.
For example, I will show you colored squares, masked by a picture, and you
will tell me whether the next set of colored squares has the same pattern.
The first component of working memory is the executive regions in the PFC.
How the PFC manages to maintain activation over an extended period appears to be
through the sorts of recurrent connections we studied in looking at neural
network modeling:
Note the role assigned to dopamine
in activating the initial cueing rule.
The next component of working memory are the sets of neurons that
provide the content to be activated by the PFC executive regions.
Note a further division of labor:
If the objects and/or relations presented by the sensory input that are to
be maintained in WM are completely new, these are being processed
by the hippocampus, which contributes to the maintenance network
If the objects and/or relations from the sensory cortex are available in
established memory networks, these are network are linked to the
executive functions in the PFC
Either way, we can tentatively answer one major question: researchers are
increasingly confident that memories are links binding together patterns
developed through the processes of perception.
That is, a memory of a house holds together the patterns that captured what
the house looked like as created in the high-order visual cortex in the ITL
This logic of binding together the networks used in perceptual processing
(which, remember, includes complex multi-modal associations) is true
for both Working Memory and episodic, semantic, and procedural
memory.
The question, then, is how is the binding done.
The short answer is, “We don’t quite know yet, but we’re working on it.”
Let’s look at a recent article that stresses the importance of the multi-modal
high-order association regions of the posterior parietal lobe, a region (with
strong connections to the insula) that is receiving ever greater attention:
The abstract reads (in part):
The hypothesis that the neural network supporting successful episodic
memory retrieval overlaps with the regions involved in episodic
encoding has garnered much interest; however, the role of the
posteromedial regions remains to be fully elucidated.… Our results
provide further evidence that posteromedial regions constitute critical
nodes in the large-scale cortical network subserving episodic memory.
The article is part of a larger argument that the posterior parietal cortex is
central to integrating all the components that comprise complex human
episodic memories and that it remains important as the site of binding for
both the process of encoding (creating) the memory and for the reactivation
of all the components on later retrieval of the memory.
The posterior parietal cortex provides the
means to bind components into memory,
but it does not do the binding itself.
This is the job of the hippocampus and its
associated cortices in the medial temporal
lobe:
Note how this model for the creation of
episodic memory looks extremely close to
that for Working Memory.
How the hippocampus does its job is a subject
of intense debate: some argue it is the site
for binding of memories; other argue that it
facilitates the binding, which happens
elsewhere.
It is clear that sleep plays a part in the process of the creation of memories
and that the hippocampus displays distinctive activation patterns during sleep
that are crucial for the consolidation of memory. This, however, is a topic for
next week.
For this week, we have one last issue: what gets remembered.
Remember the diagrams from last week:
The dopamine SEEKING system, the amygdala
FEAR system, and the other emotional networks
including the insula all signal the hippocampus.
That is, saliency is a strong component in gaining
access to the hippocampus’ memory system.
The brain tries to remember what matters.