Posted comment on ´Attention, working memory and phenomenal experience of WM content: memory levels determined by different types of top-down modulation` by J. Jacob, C. Jacobs and J.Silvanto and published in Frontiers in Psychology volume 6 Article 16033 October 2015
Jacob, Jacobs and Silvanto explored in their article the role of top-down attentional modulation of the content of working memory and concluded that the representation of the original memory in the centre of the focused attention achieved conscious awareness and this process also requires the suppression of all incoming visual information via inhibition of the early visual cortex. They began their article by defining their accepted model of working memory (model of Cowan, 1988) where working memory is seen as an activated long-term memory able to retain a number of activated representations in parallel some of which may be re-enacted long-term memory representations. They also quoted the extended model of working memory by Oberauer (2002) which suggests a store of reactivated long-term memory representations plus a capacity limited short term store (zone of direct access) and a store containing a single item linked to focused attention (FOA) which provides the content for goal-directed processing and is the only item to reach conscious awareness. The content of this working memory was said to be experienced as an image with qualia and could be scrutinised and modulated. The authors argue that reaching conscious awareness involves more than just modulation of the actual memory trace involving attention and requires in addition inhibition of visual input.
Jacob, Jacobs and Silvanto continued their article by proposing own hypothetical model relating working memory to attention. They explained that information exists in different states dependent on the level of attention and that this translates to memory. Three memory levels were proposed with relation to attentional control: one level with non-attended, non-conscious memory with no attentional modulation; second level with attended, phenomenally non-conscious memory with an enhanced actual memory trace due to attention; and the third level with attended, phenomenally conscious content with an enhanced memory trace and top-down suppression of visual input. The authors claimed that their model was distinct from previous ones because of distinct, non-conscious memories and conscious, attended memory states.
What followed was evidence indicating a dissociation between attention and phenomenal experience of memory content. Jacob, Jacobs and Silvanto began with the proposal that non-conscious items can be attended to, encoded and maintained in working memory. Research supporting that view was quoted as coming from the work by Soto et al. (2011) where subjects were able to maintain encoded information and use it later even when at the time they were unaware of the cue and distractors were present in the maintenance period. This indicated to the authors that attention was on the information held in the memory store (ie. FOA) in the delay period. Further research from Feredoes et al. (2011) showed that the working memory trace maintained in presence of distractors was brought about by the top-down facilitation of visual cortical regions maintaining the working memory content. Other brain areas were also reported as being involved in maintaining non-conscious items in working memory as changes in the right mid-lateral prefrontal cortex, orbitofrontal cortex and cerebellum were observed with activity in the dorsolateral prefrontal cortex, anterior prefrontal cortex and posterior parietal regions observed in visual working memory. Also, it was found that subliminal shapes in visual working memory guide attention and facilitate working memory performance. The authors therefore concluded that working memory can have non-conscious representations and so information in the FOA may not be necessarily conscious. This supported the view that attention may be allocated to working memory content without the content being conscious and hence, additional processes are required for a phenomenal experience to occur.
The dissociation between attended and conscious representations was reported in the article as being because once memory content has been brought to conscious awareness then it interacts differently with external input than with non-conscious memory content even if both are attended. The encoding of concurrently presented visual information could be changed by for example raising the level of the threshold at which information is detected and this could be independent of the similarity between the mental image and visual input. This was said to prevent the mental image from being weakened, but the reverse could also occur with external input impairing the conscious experience of memory content. The encoding of external stimuli matching working memory content was reported as being enhanced and reaching visual awareness more effectively whereas dissimilar information was said to be suppressed. In this way, working memory was said to act as a ´gate keeper`.
Jacob, Jacobs and Silvanto continued their discussion about their model demonstrating that consciousness and attention are dissociated and that more than just attention is required to bring content to conscious awareness by saying that their 3 state model supports well-known work on consciousness by Dehaene et. al. (2006). Dehaene et al. 2006 described information processing as being subliminal, preconscious (both considered unconscious forms) or conscious. Subliminal was defined as limited bottom-up processing due to attenuated stimulus strength potentially interacting with top-down attention. Preconscious was described as being dependent on stronger signal strength, but was limited, or had no top-down attentional modulation, but did have a potential for conscious access. Conscious processing was described as involving top-down and bottom-up activation beyond a sensory threshold and consistent with working memory models with information assigned to the FOA and activated long-term memory representations as being non-conscious. Jacob, Jacobs and Silvanto stated that Baars classic Global Workspace Theory was not a model of working memory since it described how visual input reaches awareness, but they did quote it in their article as providing evidence of the dissociation between consciousness and attention. In the Global Workspace Theory subliminal, attended information processing is short-lived and there is no attended but non-conscious content whereas in their model, non-conscious, attended representations in the working memory exist and are maintained for longer periods and can survive distractor appearances.
The article then went on to discuss the different requirements for the achievement of conscious awareness with reference to the 3 memory level model put forward by the authors, Jacob, Jacobs and Silvanto. As given above, it was stated that in the working memory model, attentional modulation of the content with conscious awareness requires an additional factor and this was proposed as the suppression of visual input from external sources via early visual cortex inhibition. This view was supported by fMRI studies where subjects carried out visual motion imagery and had were observed to have reduced early visual cortex activity, but enhanced activity in the motion-selective extrastriate region V5/MT. This was explained by the necessity of knowing the source of the content. The authors said that working memory would have to know that the conscious material was internal (the so-called Perky effect where external input is confused as part of mental imagery). They claimed that this was important so that external information is not concurrently consciously experienced as it would have a stronger neural signal than the competing conscious trace from the same item sourced from visual imagery. Therefore, when the content of the working memory needs to be brought to conscious awareness there should be a bias towards internal, reactivated information and this is achieved by inhibiting the visual input from real-time external events.
The authors concluded their article by re-stating their hypothesis that there is a relationship between phenomenal experience of content in working memory and attentional control and that conscious awareness involves the creation of a second, distinct representation (previous work) generated and by the top-down facilitation of the original memory trace and favoured to incoming external information by suppression of incoming visual input at the cortical level. They concluded their article by suggesting a possible future neuroimaging study to prove their hypothesis.
What makes this article interesting is the continued exploration of the cognitive capability, working memory. This article by Jacob, Jacobs and Silvanto links working memory with attentional modulation and conscious awareness and they propose two things: one, that the working memory is multi-facetted with facets having different attendance and awareness characteristics; and two, that the facet relating to attended, conscious information obtained from the reactivation of long-term memory relies on the suppression of incoming information from the external environment in order to reduce neural competition and this suppression is elicited by top-down attentional system modulation.
Before we can discuss their hypothesis, we have to look at and provide a neural mechanism as to what working memory is and what affect attention can have on it. My view is that working memory is not like Jacob, Jacobs and Silvanto suggest that of a ´melting pot`, a single area or brain ´splodge of activity`, but a state where multiple areas are active at the same time working on multiple ´tasks` most of which are unconscious, but at least one can reach conscious awareness. (I say at least one, because divided attention can mean that conscious awareness flits between at least two cognitive events and may appear virtually simultaneous, but from a neurochemical basis are not). Working memory is not a process, but is a ´condition` where processes can occur and these processes involve common, shared tools such as sensory input, attention and decision-making and involves information which can either come from reactivated long-term memories or from newly inputted information from the external or internal environment. The idea that multiple processes are involved is not new and well-known descriptions of the working memory (eg. that of Baddeley and Hitch, 1974) has it as a group of capabilities such as the central executive (synonymous with attention), episodic buffer, phonological loop and visuospatial pad. Jacob, Jacobs and Silvanto concentrate on only the attentional system`s contribution to function, but other tools participate as well in either the bringing of the information into the working memory state or in the information`s scrutiny or manipulation.
The informational content of the working memory can be from differing sources such as reactivated long-term memories including associated emotions and event value (Jacob, Jacobs and Silvanto`s equivalent of information with quale) as well as newly sourced material from incoming input from the external environment or internally sourced information or created material via manipulation. Whatever the source, we can assume that the time the information spends in the working memory state is dependent on the firing mechanisms in play and to an extent, its relative importance to the individual whether from task performance or personal value. For example, information said to be in the short-term memory store according to Jacob, Jacobs and Silvanto is from the neuronal firing of relevant cells and therefore, the period of activation would be dictated by the time the neuronal cells could fire before chemically being exhausted and they shift into their refractory periods to replenish. The information would then fade. Repetition or manipulation of material would lead to sustained neuronal cell firing and the holding of that information in the working memory state for longer periods just like in the case of formation of long-term memories.
In neurochemical terms, the firing of neuronal cells depends on the source of the material eg. visual working memory activates different brain areas to language and this allows the scope of the working memory state to broaden if multiple skills and senses are involved in the content. The cells themselves are considered to be multi-tasking (Messenger) which allows cells to be representatives of information, but also susceptible and instigators to tools such as attentional modulation and the whole state representing an event depends on connectivity of multiple areas so that items are inputted, maintained and manipulated. Visual working memory is said to involve many areas such as the prefrontal cortex, cingulate cortex, hippocampus and entorhinal cortex (for relaying the signals, synchronicity and binding), fornix and thalamus (for basic sensory information relays), V4 and medial temporal lobe and inferotemporal lobe (for visual pathways and visual attention) as well as the cerebellum (for procedural memory and movements). Manipulation and holding of material is the responsibility of the lateral- and post-parietal cortex.
Whether the informational content of the working memory state has conscious awareness or not depends on several different factors and is independent of the source (eg. internal or external) or type of information (eg. visual, auditory). Jacob, Jacobs and Silvanto hypothesise 3 working memory levels dependent on whether material is attended or not, or conscious or not. What selects information for conscious awareness is essentially the strength of firing and strength of firing is dependent on quantity of information (ie. the more cells active, the greater the chance of reaching conscious awareness) and quality (ie. the more characteristics available including emotional status and value, the greater the chance of reaching conscious awareness) and the task at hand (ie. the more difficult the task or more complex for example the greater the chance of conscious awareness). In the case of, for example, the simple recall of a procedural memory like riding a bike then this is unlikely to evoke conscious awareness especially if other more challenging input is available at the same time, but it will and will enter the working memory state if it is coupled with learning to ride a new bike with a different gearing system. Therefore, conscious awareness of one facet of the working memory will reflect the strength of firing of the information independent of its source. However, it should be remembered that conscious awareness represents only one ´draft` of an experience if the multiple drafts theory for consciousness is to be believed. This can be compared to working memory which can also be considered as one ´draft` of information with manipulation, the addition of supplementary information (associated with ´filling in` of consciousness) providing the other ´drafts`. The link is also supported by looking at the brain areas involved. Consciousness involves the firing of particular brain areas such as prefrontal cortex, cingulate cortex and parietal cortex and these are as shown above the same areas as those said to be involved in working memory. Therefore, the two capabilities can be said to be linked although in most cases dissociated.
So, we have looked at what working memory is and how conscious awareness of some of its content can be brought about, but Jacob, Jacobs and Silvanto expanded this by hypothesising that working memory and consciousness relies in 2 cases of their 3 memory model on the involvement of the attentional system. Only the non-attended, non-conscious form had no attentional modulation according to them and this can be is explained by considering this form of working memory as being associated if at all with recall of memories without the need for any further processing. The only disadvantage of this definition is that it is unlikely that in this case the working memory state at all since the working memory state is usually associated with information manipulation and pure recall does not require any further processing to be totally effective. One of the other forms of memory described by Jacob, Jacobs and Silvanto states that information is attended, but phenomenally unconscious and so attentional modulation can lead to enhancement ie. more processing, but still remains under the threshold for conscious awareness. It is only the final form where conscious awareness is attributed to working memory content and this attended information is in the centre of attentional focus and can be scrutinised and modulated. The link between attention and consciousness is not new with Dehaene and Changeaux`s 2011 Neuronal Global Workspace Theory proposing that attention acts as a selecting mechanism for conscious contents and working memory as a specific store.
Although we think of attention as a discernible force, biochemically it is not. It is a mechanism that instigates the strengthening of certain information and the weakening of other rather like a dial so that some information is attended and other not. This is an important quality if we want a working memory state where some of the information active at that time can be scrutinised and manipulated and the rest ignored or simply carries on. Competitive selection of information can be based on feature strength or even biased because of stimulus colour, movement or emotional value. The level of attention can also vary with a low level described by Marchetti as being with or without consciousness or a high level as associated with selected events, manipulation and decision-making for example. It can also be bottom-up based on the stimulus features and lower level sensory pathways and automatic recall of values, memories for example, or top-down meaning that the allocation of the resources is under the control of the higher cortical areas and dependent on memory, values, associations and decision-making. Independent of which attentional system is exerting its control, the first 270 milliseconds of any event is neurochemically the same and it is only after this time that the allocation of further deciding attentional resources occurs. This early period could be described as the preconscious period for some events with them either reaching conscious awareness later if focused attention is applied (after 300 milliseconds) or if no attentional resources are directed at them being the ´never-conscious`. Attention may not also be considered as a single focused capability centred on a limited area since Marchetti also described attention as being ´diffuse` ie. like ´gist`, spread over a number of aspects of one event. In Jacob, Jacobs and Silvanto`s model the focus of attention is the conscious event and the diffuse attentional state produces no conscious awareness which differs from Marchetti and others who say that even in this condition, conscious awareness can occur. This is credible if we consider diffuse attention as ´gist` – we may not be exactly aware of all facts, but we have an overall understanding. Therefore, linking information in working memory and how some of this reaches conscious awareness and some not relative to the amount of attentional resources aimed at it is understandable.
However, Jacob, Jacobs and Silvanto went further with their working memory model by saying that the attentional system inhibited certain informational input into the working memory state if the informational content was of a particular kind ie. was information obtained from reactivated long-term memories for the same event as being observed in real-time. This inhibition was brought about to reduce the competition for working memory capability from the incoming information from the environment which they said would produce naturally stronger neuronal firing and therefore, evoke conscious awareness in preference to the reactivated familiar material. Two copies of the same material would exist with the copy of the newly inputted material being stronger. Therefore, Jacob, Jacobs and Silvanto´s inhibition hypothesis is understandable since: competition would increase the perceptual load and therefore, certain characteristics may be ignored; it provides a reason why material is not processed again if it has already been processed and stored (eg forms capability of object recognition); and the reactivated information may have more recalled associated material with it such as emotional state and personal value than the new input. The hypothesis is supported by the observation that we are often unaware of a change in a person`s appearance for example (attentional blindess). This implies that our recognition of that person relies on recall of stored information in response to cues from the real-time encounter, but close examination of the person`s appearance in real-time does not occur. Although this may be negative in that changes are not observed, using recalled information has the advantage that it is more than just visual characteristics for example and that other information and associations are also part of the reactivated information and hence, the capability of the working memory to process material is strengthened and so is also the chance to reach conscious awareness. However, this inhibition may not be possible as already indicated if the reactivated image is too different from the incoming information and may be detrimental since no updating of the stored information from sensory information will be possible.
The mechanism hypothesised by Jacob, Jacobs and Silvanto as bringing about the inhibition was top-down attentional control at the level of the early visual cortex V1, but herein lies a problem. Jacob, Jacobs and Silvanto compared the informational content in the working memory of the reactivated memory as visual imagery and visual imagery is associated with activation of the V1. Therefore, suppression of activity in this brain area by the attentional system would automatically inhibit the working memory state and suppress the visual information being recalled. Possible explanations to explain this discrepancy if this hypothesis is correct are: that the neurons firing in response to input in the V1 are not the same ones of the neuronal cell assembly group representing the stored image; or that the inhibition occurs at a lower level than the V1 cell hierarchy so that visual features are not discernible. Further research into visual imagery particularly using real-time neuroimaging is required to explore the capability.
Therefore to conclude, Jacob, Jacobs and Silvanto`s model for working memory and the levels of information relating to attendance and conscious awareness has merits in that it provides support for the view that working memory is not a single item capability, but is multi-facetted with each facet having different characteristics regarding informational content, conscious awareness and attentional system involvement. In the case of conscious awareness of one type of working memory content, that of reactivated visual images, then Jacob, Jacobs and Silvanto`s model proposes that top-down attentional modulation occurs that inhibits the input of real-time visual information if the event being observed is the same as that of the reactivated store. This has the advantages that competition for cognitive resources from the stronger real-time image is removed and that associated information such as emotional worth is also recalled and is available for processing and manipulation in addition to the event features. This strengthens the firing and is likely to increases the chance of bringing the content to conscious awareness. However, inhibition of visual input also has the disadvantage that the updating of stored information from real-time external events is prevented. Inhibition of this type could be detrimental for those people that suffer from memory recall problems since the information reactivated may be sufficient enough to trigger inhibition, but not substantial enough to aid working memory performance. Therefore, the topic deserves further investigation.
Since we`re talking about the topic………………..
…..can we assume that the same type of inhibition can be observed with auditory memories under the same conditions and also that the same type of inhibition occurs when the events are multi-sensory?
…..if the experiments are repeated using the same reactivated event as the external event being experienced, at what level of dissimilarity of content is the inhibition of the external input removed? Does this correlate time-wise to when the subject becomes conscious that a change has occurred? Does this occur even if that change refers only in a change in emotional worth?
……the administration of ketamine leads to an increase in irrelevant information in working memory. Therefore, if ketamine is given and the experiments repeated, would conscious awareness be on multiple events akin to divided attention and how would this affect the suppression of real-time visual input? Is it possible that attention is diffuse rather than focused, or is it focused on multiple events (like divided attention) and hence, not reach the threshold value for visual input inhibition?
… what effect could working memory training have on the inhibition of V1 activity?