Posted comment on ´The contribution of AMPA and NMDA receptors to persistent firing in the dorsolateral prefrontal cortex in working memory` written by B. Van Vugt, T. Van Kerkoerle, D. Vartak and P.R. Roeslfsema and published in Journal of Neuroscience 2020 40(12) p. 2458 doi 10.1523/JNEUROSCI.2121-19.2020
SUMMARY
Van Vugt and team report in their article the results of their investigation into the contributions played by the different glutamate receptors, AMPARs and NMDARs, of the dorsal lateral prefrontal cortex (DLPFC) in the persistent firing observed during a working memory task. The current theory says that NMDARs because of their long time constants play a greater role than AMPARs which have shorter time constants. Van Vugt and colleagues tested the hypothesis that in the oculomotor delayed response task (ODR) the NMDARs maintained the information associated with the cue as a pattern of persistent neuronal firing during the memory delay whereas the AMPARs activated the neurons during the sensory stimulus and persistent firing phase. Hence, they found that both NMDARs and AMPARs contributed to the persistent activity observed.
For their experiments, three male macaque monkeys performed an oculomotor delayed response task (ODR). The monkeys were first trained with a fixation point of a red circle on a grey background and had to gaze at a fixation window centred on the fixation point. After 300ms they were then presented with a visual cue of a white circle at either the appropriate neuron`s receptive field (RF) or at what was termed the anti-preferred location which was the fixation point`s mirrored location. The visual cue was turned off after 150ms, but the monkey had to maintain the fixation window for another 100ms before the fixation it was extinguished. This was the signal for the monkey to make a memory guided eye movement back to the target window which was centred on the location of the previous visual cue. Correct responses were rewarded with apple juice. Trials where the monkeys failed to maintained fixation before the fixation point was extinguished were stopped. All stimuli were generated by computer software and eye movements were recorded with a video eye-tracker. Iontophoresis was used to eject small amounts of glutamate receptor antagonists (APV for NMDARs and CNQX for AMPARs), which were enough to disrupt, but not abolish neuronal activity.
Neural activity was recorded from the dorsolateral prefrontal cortex (DLPFC) including frontal eye fields and surrounding cortex. Only selected neurons that showed sustained firing were used. Three blocks of approx. eighty trials were recorded: first without drug delivery by maintaining holding current (pre-drug recordings); a recording block when drugs were administered by applying the ejection current (during drug recordings); and finally a recording block without drug delivery, but again with the holding current maintained (post-drug recordings). These phases began at different times after the effect of the antagonist drug produced noticeable spiking activity.
Data analyses were performed on the results obtained. The ODR task was divided into two time periods: spontaneous activity (lasted from 300ms before the stimulus onset up to stimulus onset); and task-related activity (from the stimulus onset up to saccade onset, with ´saccade` intended here as equivalent to the neural firing pattern formation). Values calculated included cue-driven activity in the time window from 50-250ms, persistent activity from 300-1150ms after cue onset (starting 150ms after cue onset) and saccade related activity from 200ms before the onset of saccade. Spatial selectivity for each individual cell was quantified using the measurement d´ which showed how well a single neuron could distinguish between the memories formed for the two locations. Statistical analyses were performed including two-sided t tests and three-way repeated measures ANOVA. The authors also performed a stratification analysis to see if APV and CNQX differences were due to relatively low firing rates in cells when dealing with anti-preferred locations and the results led to only certain cells being used in the analyses.
The results of Van Vugt and team`s experiments showed that overall performance on the task was high. Small doses of the NMDAR antagonist, APV, applied so as to perturb activity, but not to abolish it produced no consistent effects on performance accuracy. Small doses of the AMPAR antagonist, CNQX, when applied again like APV so as to perturb activity, but not to abolish it, produced instead small increases and decreases in performance accuracy. These were interpreted by the authors as possible changes in the motivation of the animals to perform the tests under these conditions.
The sets of experiments carried out involved the recording of neural activities of single DLPFC neurons during the ODR task. It was found that most neurons exhibited persistent firing during the memory periods. Neurons showed increased firing during the full duration of the trial when the visual cue was presented at the preferred location of their receptive fields (RFs) and a suppression of firing when the visual cue was presented at the anti-preferred location. The iontophoretic application of APV (the NMDAR antagonist) suppressed baseline activity before the visual cue onset as well as spiking activity in both monkeys at the preferred location during visual cue responses, during the persistent activity phase (time window 300-1150 ms) and the saccade window (200ms before the saccade). The suppression for the preferred location responses was greater than for the anti-preferred locations. Therefore, the antagonism of the NMDARs was shown to weaken spatial selectivity. Computed d´ measurements of how well a single neuron could distinguish between the memories for the two locations showed that the values decreased in both monkey subjects. The cessation of APV administration led to a return of spiking values, although not to the original values and neither was it observed for both monkeys.
Van Vugt and colleagues looked at the time course of the APV suppression effect by plotting the difference between the spiking activity before and during its administration. For the preferred location, the main effect of APV was found to be on the firing rate, but there were no significant interactions between the drug effect and the different test periods (eg. spontaneous, visual, delay and saccade activity periods). The decrease of the cue-driven response supported the hypothesis of a general multiplicative effect of NMDARs on spiking activity, but appeared not to support the hypothesis that the NMDARs have a specific role in the generation of persistent activity. Investigation of small subsets of neurons with a visual response without the delay activity found that APV suppressed the visually driven activity of these neurons supporting the view that NMDARs play a more general role in both cue-driven and persistent activity.
In the case of the iontophoretic administration of CNQX (the antagonist of AMPARs) Van Vugt and team found that there was a decreased level of persistent firing associated with the memories of neurons for both the preferred and anti-preferred spatial locations. The baseline spiking activity before the visual cue onset was also suppressed as well as the spiking activity in the cue window, memory window when the visual cue was presented in the preferred location and saccade window. When the visual cue was at the anti-preferred location then significant decreases of activity were observed in the cue window, memory window and saccade window in one monkey, but not in the other. The decrease in d´ (performance of a single neuron for each location) was observed with both monkeys to different degrees and there was no effect on the Fano factor (defined here as the variance of firing divided by the mean spiking activity averaged within a specified time window). Cessation of the iontophoretic administration of CNQX did not restore the spiking activity to pre-drug levels and with some neurons no trend of recovery was even observed. Therefore, the authors concluded that CNQX administration induces long-lasting effects. The influence on firing rates was found to be significant for the preferred location, but there were no interaction effects between the drug and time period indicating that the drug effects were similar across the different time periods.
Comparison of the effect of APV and CNQX on delay activity showed that the decrease in d´ was greater for NMDAR antagonist administration than for AMPAR antagonistic administration. No definitive conclusions were made by the authors because of the differences in ejection currents, poor controls of factors such as efficiencies of drugs, distance between neuron and pipette, diffusion and clearance of drugs for example. Therefore, the authors determined how well the influence on delay activity for the preferred cue location predicted the influence on delay activity for the non-preferred location. The authors found that in the case of APV administration the value was not significant and therefore, it was concluded that the decreased activity in the preferred direction was not a good predictor for the activity decrease in the anti-preferred direction. This interaction was not seen with CNQX where the value was significant and therefore, the prediction statement was valid. The difference between the two antagonists was also found to be significant. This was interpreted as the two likely demonstrating the distinct actions of the NMDARs and AMPARs. The authors explained that when glutamate binds to the AMPAR then the associated ion channel opens and the cell is activated. However, when glutamate binds to the NMDARs, the associated ion channels and receptors are normally non-functional due to the binding of magnesium, which has to be removed before the ion channel functioning occurs. This could explain the observation that a decrease in delay activity in the preferred direction was not always accompanied by a comparable decrease in the anti-preferred location since in this case the receptors may still be blocked because of the lower neuronal firing rate associated with the non-preferred location. Therefore, the administered APV could not exert its full effect.
Van Vugt and team concluded that both AMPARs and NMDARs contribute to persistent activity during working memory tasks. The administration of APV, the NMDAR antagonist, led to a decrease in persistent firing associated with those active neurons representing the preferred spatial location in short-term memory, but had little effect on the neural representation of the anti-preferred location. Therefore, APV decreased the information conveyed by persistent firing about the memorised location. The administration of CNQX, AMPAR antagonist, gave a different effect with decreased persistent firing associated with the actively firing neurons in the formation of short-term memories for both the preferred and anti-preferred spatial locations with the baseline spiking activity suppressed even before visual cue onset. Van Vugt and colleagues stated that their study could provide a base for further studies on the effects of excitation of the dorsolateral prefrontal cortex area and demonstrate its connectivity to other brain areas as well as explaining how task-relevant information is kept online during memory recall delays.
COMMENT
What makes this paper interesting is that it looks at the roles of NMDARs and AMPARs in the dorsolateral prefrontal cortex (DLPFC) during the transient world of working memory involving information processing and short-term memory. The general experimental set-up that of the oculomotor delayed response task (the ODR) determines like any other experimentation what systems are required for the successful performance of the task set. The ODR used by Van Vugt and team in their experiments was basically a conditioning task. The monkeys were trained to look at visual cues and perform eye movements in response to changing locations. Failure was determined by the monkey not performing the learnt eye movement at the correct time.
Each stage of the experimental set-up of this ODR task required a number of sensory and cognitive processes. The first stage, that of training to gaze at a visual cue (the red circle on a grey background) at a particular location (fixation window centred on the fixation point) on a screen, required at least the application of the visual input system and attentional system. Short-term memory formation of the visual cue`s location was also required in order that the task could be performed successfully later. The type of memory employed is probably spatial memory and not particularly visual memory since the red/white colours of the cues were essentially immaterial to test success since the monkey only needed to know about the location of the lighted point. Then, after 300ms the monkey was presented with another visual cue (this time white) for a time period of 150ms at a location either at the same point as the first (the preferred location) or at its mirrored location (termed the anti-preferred location). For this stage, a number of sensory and cognitive capabilities were in play, eg. the visual system (looking at the visual cue), attention system (shifting attention to the cue, maintaining attention on the visual cue and avoiding distraction), motor system (shifting the gaze) and short-term memory of this position (spatial memory) whilst also remembering the location of initial target (working memory and short-term memory). The monkey had to keep its gaze on the location first for the time period of 150ms when the visual cue was lit and for a further 100ms once the visual cue was extinguished guided by the placement of the fixation window. This 100ms period provided the ´delay` where both neural representations in the form of short-term memories had to be maintained in the absence of the visual test stimuli. (Although, it could be said that the fixation window also produced a visual signature and real-time sensory systems relating to visual information were still being stimulated.) The extinguishing of the fixation window was the conditioned signal for the monkey to move its gaze back to the initial location (termed ´saccade` by the Van Vugt team) and this stage employed a number of different systems including motor movements and the employment of spatial memory and comparisons of temporarily stored information. The success of the experiment was the shift of gaze back to the preferred location and failure was effectively any other eye movements (ignored completely as test results) or the movement to the anti-preferred location. Therefore, the experimental set-up required overall a number of different sensory and cognitive systems (eg. visual system and visual memory, motor systems and motor memory – head and eye movements, attentional system, working memory – information processing and short-term memory as well as spatial memory) and since the DLPFC area was used as the test area, results of Van Vugt`s experiments should indicate the role that this particular area plays in this type of task and more specifically, what roles the two glutamate receptor types have in this particular task.
What Van Vugt and colleagues found from their ODR task based experiments was that both AMPARs and NMDARs of the DLPFC area contribute to the persistent neuronal activity required for successful working memory employment and task completion. Contribution of the two sets of receptors appeared not to be the same. The administration of APV, the NMDAR antagonist, produced decreased persistent firing associated with the firing of neurons required to produce the neuronal representation of the preferred spatial location in short-term memory, but had little effect on the neural representation of the anti-preferred location. Therefore, it was said that the APV decreased the information conveyed by persistent firing about the memorised location required. This indicates that NMDARs are essential for the short-term memory formation of the relevant information. CNQX, the AMPAR antagonist, however, showed a different effect since decreased persistent firing was observed associated with the short-term memories for both the preferred and anti-preferred spatial locations with the baseline spiking activity suppressed even before visual cue onset. So, how does this fit in with what we know about the DLPFC area and NMDAR and AMPAR contributions associated with the task of working memory (information processing) and short-term memory formation?
As far as working memory goes, there are different types eg. visual, tactile and spatial which all share the main aims of information processing and the holding of multiple simultaneous informational units. Working memory should perhaps be considered more of a ´state` allowing these functions to occur and therefore, the requirements of the system relate to information in and out, maintenance of information and assessment of information (whether internally within the active state or sent to somewhere else for assessment). The location of the active ´state` plus connectivity of this location to other areas is likely to follow the function. Therefore, separate senses have been reported as having differing working memory locations (eg. visual – infero-temporal area, V4 and medial temporal area; tactile – S1,S2; and auditory where even time, frequency and intensity of sound are all located to separate modules – work by Pasternak). The location of working memory and its connectivity appears to be traditionally located to the prefrontal cortex following the connected pathway to the anterior cingulate cortex, then hippocampus followed by the anterior cingulate cortex, but other areas also appear to be involved. For example, the lateral intraparietal cortex (Takeichi – working memory training led to increased myelination in white matter neurons of the intra parietal sulcus and anterior corpus callosum), plus areas associated with attention (guided eye movements, saccades), manipulation of information (post-parietal cortex) and item maintenance (post-parietal cortex and lateral occipital cortex).
So, what roles does the DLPFC play in working memory? It is likely that the DLPFC is associated with executive control, the maintenance of items and value assessment that are in play in the particular working memory state being considered. The executive control function which is required in the conditioning task appears to be a common function of lateral prefrontal cortical areas (D`Espisito, Macdonald) and can even demonstrate differences in function according to hemisphere (eg. D`Espisito – right hemisphere, plan generation; left hemisphere, plan execution). Within the DLPFC itself, flexible neuronal tuning supports top-down modulation of task-relevant processes and neuroimaging has shown that the DLPFC can carry out cognitive control adjustments based on the detection of conflict. This could be because of its connectivity to the anterior cingulate cortex, an area known to monitor for conflict.
With regards to spatial working memory, most researchers support the view that the DLPFC is essential for this type of working memory whereas its ventral lateral counterpart is required for the processing of language, communication signals such as facial expressions and audiovisual working memory (Plakke). However, studies to transfer this function to humans appear to show that spatial working memory (as required in Van Vugt`s ODR task) in the human requires not only the DLPFC, but also the precentral sulcus. Part of the working memory function is the requirement to maintain the neural representation of the item or items within the working memory state so that the required processing of the information can be carried out. This appears to be a function associated with the general prefrontal cortex area (Baier). In the ODR task, two neural representations have to be maintained – two visual cues (one red, one white) which have two locations (one preferred, one anti-preferred). It has been found that working memory can hold simultaneously three items decreasing as the number of items stored is increased (Anderson). Therefore, the two pieces of information relating to the ODR task undertaken by Van Vugt`s subjects, bearing in mind that one piece of neural information with regards to spatial memory has feature plus location characteristics, is well within the capability of the working memory state. The quantity is attenuated by the content overlapping (visual cues may only differ by colour) and both being goal-relevant (Soto).
The third function of the DLPFC and the working memory state is the assignment and comparison of ´value`. In the ODR task, the ´value` is deemed according to location, ie. preferred location versus anti-preferred location. The first brings reward, whereas the second (as well as no movement) does not. (As said above, the colour of the visual cue is probably immaterial and it is only the location and relevant eye movement that is counted as success.) Recognition for this type of function relating to this area is well known, eg. increased activity in the DLPFC when actions are selected and initiated (Spencer); neuronal representation of information relevant for credit assignment reported in the DLPFC with specifically, the neuronal activity reflecting both the relevant cues and outcomes at the time of feedback and were stable over time (Asaad); and DLPFC activity denotes comparison of two strategy values (Wan). Value is to some extent recorded since the monkey knows that one particular movement instigates on retrieval, reward. This value is assigned and recorded during the training period and those physiological changes associated with long-term memory are reflected in the control values of the NMDARs and AMPARs at the beginning of the test periods. This value assignment is likely to involve the connectivity of the DLPFC with the anterior cingulate cortex (ACC). It was found that a larger proportion of neurons were activated in low motivational conditions in the DLPFC than in the ACC and the onset of this activity was significantly earlier in this region (Amemori). Therefore, this indicated that motivation and value judgement required firing of both the DLPFC and ACC, but to different degrees and at differing timings.
The ability of the DLPFC to perform these functions relating to working memory is dependent on the area`s physiology and connectivity. From a connectivity perspective, it is known that the DLPFC is part of a general network of frontal, parietal and insular brain regions activated in response to a wide range of demanding task conditions (Brosman). In particular, there is competition between the DLPFC, PFC (there are reciprocal connections to the orbitofrontal cortical area OFC) and striatal regions (Daw). With reference to spatial attention, cortico-striatal pathways are particularly important with the identification two bilateral convergence zones (in the caudate and putamen) receiving input from the OFC, DLPFC and parietal regions.
The functionality of the area also relies on its neuronal firing capability and relating to this, Van Vugt and team explored the capabilities of the two glutamate receptors, NMDAR and AMPAR located within the DLPFC. It has been shown that the PFC contains neurons that can perform multiple tasks simultaneously in the form of both working memory and attention functions (Messenger). This would be a valid explanation of how neuronal activity is controlled for informational processing demands. From a neurotransmitter perspective, working memory is normally associated with the neurotransmitters, dopamine and GABA. It has been reported that there is dopamine connectivity between the left and right fronto-parietal brain networks (Cassidy) that may adapt flexibly to cognitive demand. Poor working memory performance was related to deficient cortical dopamine release. However, with regards to amphetamine, a U-shaped functional modulation of working memory performance and the medial PFC effect was observed (Lapish) where moderate increases in monoamine efflux would increase attractor stability, but high frontal levels would diminish it. The role of GABA in working memory is clearer with a report showing that GABAergic dysfunction in the PFC observed with increasing age is linked to decreased working memory capability (Banuals) and in particular with the DLPFC that lower GABA elicits a greater loss of performance particularly with high task demands (Yoon).
However, Van Vugt and team`s article concentrated on the roles of the two glutamate receptors, NMDAR and AMPAR on working memory function. In general, there are known stages to glutamate functioning at the neuronal level. The first is the binding of the neurotransmitter to the receptor protein eliciting the activation of its associated G protein. This is followed by the activation of the effector systems through the G protein signal cascade and finally, the post-synaptic neuronal functioning. In the case of the NMDAR, bound magnesium ions are released on arrival of the action potential on cell firing and neurotransmitter binding and this release causes the associated calcium channel to open. This calcium ion permeability of the neuronal NMDAR is under the control of the cAMP protein kinase A signaling cascade of the post-synaptic area (Skerbidis, Nicolls). Some NMDARs however, are associated with sodium channels like the AMPAR and these are then opened when glutamate binds. Others are associated with SK channels and potassium ion channels causing an influx of potassium ions. Binding of glutamate released by the action potential reaching the presynaptic area can also occur to the post-synaptic AMPARs and these are associated mainly with sodium channels, although some are linked to calcium channels. Both lead to the appropriate ion influx. If firing is sustained then long term potentiation (LTP) may occur. LTP occurs primarily because of long-term AMPAR changes causing increased sensitivity of the neuron to the firing stimulus. There are an increased number of AMPARs at the membrane surface resulting in higher sensitivity and strength of firing to stimulus. The alterations in AMPAR functional properties are coupled to trafficking, cytoskeletal dynamics and local protein synthesis (Derkach – PIP3 turnover required at synapse to maintain the clustering of AMPAR; Bacaj – AMPAR nanodomains are often, but not systematically, colocalized with clusters of the scaffold protein PSD95).
Therefore, the main difference between the NMDAR and AMPAR is their response to sustained firing – a condition required in the formation of memories. With regards to Van Vugt`s experiments, long-term physiological changes are likely to have occurred in the training periods before the ODR test phase. Those training periods would have led to LTP of the relevant neurons meaning that insertion of AMPARs in the post-synaptic membrane would have occurred (Derkath, Plant, Ehninger, Rauer). The LTP would result in increased synaptic strength and increased susceptibility to depolarization for neurons associated with the visual cues, but from an experimental perspective the changes would be reflected by the higher control values compared to non-trained animals. LTP is not always observed in sustained firing conditions and sometimes long-term depression (LTD) results. This means that there is increased sensitivity of the neuron to inhibit firing on input. However, LTD is not likely to have occurred here since the DLPFC neurons were associated with excitation and it should be remembered that any changes would have been observed and accounted for by the control.
The role of glutamate receptors in working memory is well known. For example, NMDARs containing the NR2A subunit in the PFC (particularly the layer 2/3 pyramidal neurons) were found to be required (McQuail) and the level of task irrelevant information is affected by NMDAR antagonists (Cage). Van Vugt and team`s experiments investigated the roles of both NMDARs and AMPARs in the DLPFC during a task that required working memory. They found that the AMPARs and NMDARs contributed to the persistent neuronal activity in the DLPFC during their working memory tasks. The antagonist of NMDARs, APV, decreased persistent firing associated with the memory of neurons firing to represent the preferred spatial location, but had little effect on the neural representation of the anti-preferred location. Therefore, the APV decreased the information conveyed by persistent firing about the memorised preferred location.
The antagonist of the AMPARs, CNQX, showed a different effect with decreased persistent firing associated with the memories of the firing neurons for both the preferred and anti-preferred spatial locations with the baseline spiking activity suppressed even before visual cue onset. Therefore, Van Vugt and team concluded that the receptors provided different contributions with NMDARs showing a general multiplicative effect on spiking activity with no specific role in the generation of persistent activity, but a specific role in visually driven activity of those neurons supporting the desired cue location, whereas the AMPARs were associated with the memories of neurons for both the preferred and anti-preferred spatial locations. This is supported to some extent by work on the NMDARs of the lateral PFC again of macaque monkeys during a working memory task (Ma). Here it was found that that acute injections of the NMDAR antagonist, ketamine, both weakened the rule signal across all delay periods and amplified the trial-to-trial variance in neural activities (i.e., noise), both within individual neurons and at the neuronal assembly level. This resulted in impaired working memory performance. In the minority of post-injection trials when the animals responded correctly, the preservation of the signal strength during the delay periods was predictive of their subsequent success. These findings suggested that the NMDA receptor function would be critical for establishing the optimal signal-to-noise ratio in information representation by assemblies of PFC neurons. This supported Van Vugt`s study where application of APV reduced neuronal spiking and this was also reflected by the success of the memory being maintained so that the eye gaze moved back to the preferred location at the correct time.
The results of Van Vugt`s experiments can also lead us to make further observations about the actions of both NMDARs and AMPARs of the DLPFC during this type of task. Van Vugt termed the firing of DLPFC cells responsive to spatial information as ´delay cells`. These neuronal cells are activated by the visual cue and remain active during the period between it being extinguished and the time when the gaze shifts back to the preferred location. This is termed as the ´memory delay`. In neurochemical terms, this would accounted for by the formation of sensory stores relating to the visual cue and then maintenance of these firing neurons in a short-term memory store representing their neuronal firing pattern. This would be held in the working memory state whilst the distracting task of the alternate visual cue is presented and extinguished. Van Vugt and team concluded that there was a subset of the firing population which was activated by the stimulus, but switched off when the stimulus was no longer visible and a set of ´delay cells` that exhibited persistent activity. They suggested that this persistent firing could result in reverberatory excitation within the area and between cortical areas eg. between the cortex and subcortical structures including the thalamus and cerebellum.
So, the first question is whether subsets of neurotransmitter receptor populations within an area exist or not? The answer is affirmative. We know that glutamate is a single neurotransmitter which can bind to and activate different receptors and multiple forms of receptors. Even in the case of the NMDARs, some are linked to sodium channels, others potassium channels for example. Receptors can also have different characteristics and functions dependent on their locations within an area, eg. TANS. In the experiments described by Van Vugt, the NMDAR populations respond to transient visual cue appearance as well as maintaining the neural representation in its absence. Therefore, it is likely that it does not require the whole population to maintain the stimulus`s characteristics and selectivity could depend on the type of channel associated with it or even on the receptors location. Van Vugt and team hypothesised that with NMDAR involvement, membrane depolarisation leads to neuronal firing with strong effects on neurons driven by the stimulus and smaller effects for weakly activated cells. In persistent firing, ie. that maintaining the neural representation of the visual cue at the preferred location, then Van Vugt says that glutamate needs to bind and the neuron has to be depolarised to release the magnesium ions from the NMDAR channel to unlock the block which exists at rest. In this case, it is likely that the NMDARs referred to here are SK channel linked and respond with an influx of potassium ions. If this is the case, then firing is abolished and recovery of the cell instigated. Therefore, these neurons could represent those responding to the presence of the actual visual cue whereas the other firing subset would maintain the neural representation in the working memory state (essentially the short-term memory) whilst the distracting cue is present. This is because at one time period of the experiment, two sets of active neurons exist: one maintaining the preferred location in the working memory and the other the real-time visual stimulus. Therefore, the functions of the NMDAR populations could be temporarily split. This has been observed in the premotor cortex where two targets representing two movements of the sequence are represented in the working memory by two subpopulations of the neurons (Sanechi). Also, with more relevance to the experiments of Van Vugt and team, by findings from Andersen, who found that spatial attention can be divided effectively between separate locations whereas non-spatial features have a so-called ´global effect` whereby items having the attended features may be preferentially processed throughout the entire visual field. Therefore, when cues are of the same colour then even if spatially independent then the firing for both is improved since one helps the other rather than when different and competition occurs. Task similarity, task difficulty, practice, training, age and anxiety are all observed to influence the quality and quantity of information in divided attention conditions.
In the experiments of Van Vugt, the working memory content refers in general to one task-relevant group (the visual cue at the preferred location) and one task irrelevant (a visual cue at the anti-preferred location). The division goes further by the task relevant being associated with reward and therefore, firing is likely to be more robust. This was demonstrated by Donahue who showed that neurons in the DLPFC only encoded task-relevant memory signals with their congruent choice signals and these were more robustly encoded following rewarded outcomes. This is supported by work by Suzuki who says that the DLPFC performance closely correlates to the suppression of distractor stimuli and therefore, mediates selection of information.
However, Van Vugt`s experiments also looked at the contribution of AMPARs of the DLPFC in their working memory task. These were also shown to play a role in persistent activity independent of visual cue location. This demonstrates a likely general firing response to calcium ion influx mediated by the action potential. Van Vugt and team state that the AMPARs always depolarise the post-synaptic neurons in an additive manner. Therefore, NMDAR activity is induced by sensory input and NMDAR channels and ion flow occurs resulting in the activation of the AMPARs leading to persistent activity from sufficient depolarisation even in the absence of stimulus. This predominance of the NMDAR effect is supported by others. It has been shown that regarding persistent firing in the DLPFC, NMDAR antagonists almost abolished persistent activity whereas AMPAR antagonists were weak during the start of the delay period and stronger towards the end of the delay period indicating that NMDARs had a role in working memory (Wang). However, it was also said that it could demonstrate a difference in the efficacy of the receptor antagonists since the NMDAR antagonist produced stronger effects than the AMPAR antagonist in all experimental periods.
The second question raised as a result of the differences in glutamate receptor contributions is whether the persistent activity observed is due to differing contributions of and to connectivity between the DLPFC and other areas. We have already seen that working memory requires the connectivity between this area and others. In particular, the DLPFC and OFC demonstrate strong reciprocal connectivity. Connectivity is important since working memory requires an active attentional system and this is dependent on activity of multiple brain areas. The attentional system already described above as playing a role in the priority of task relevant and task irrelevant information in working memory, also provides informational selection and updating processes on the basis of strength of neuronal firing. In the case of information relevancy, for example working memory allows individuals to pay only three quarters of the maximum level of attention to relevant stimuli and to ignore unwanted stimuli and for this, connectivity of areas such as the PFC and globus pallidus are said to be important (Klingsberg). It is known that the selection process retrieving the relevant item requires the brain areas of the superior frontal gyrus, post-cingulate cortex and precuneus (Blauracke, Bledowski). Also the parieto-medial-temporal connectivity was found to be important when working memory benefits attention by strategic control when the contents overlap in goal relevant stimulus features (Soto). Therefore, connectivity to other areas is important to working memory performance and since some NMDAR populations are connected to ion channels that initiate post-synaptic signal transmission and others not, this may provide another reason for subset firing differences.
Therefore, we have established an association between working memory performance and NMDAR and AMPAR populations of the DLPFC. This is further supported by observed changes to working memory performance with reported DLPFC effects. For example, specialized cognitive training has been shown to increase working memory performance and this has been linked to the volume of the rostral part of the left DLPFC being able to predict an individual`s reponse to training (Verghese). Also transcranial stimulation (tDCS) of the DLPFC has been shown to prevent the deficits observed in working memory with stress (visuospatial worse than verbal – Bogdanov). In this case, stress was shown to impair working memory performance by decreasing the activity of the area which could be alleviated by anodal tDCS. The administration of oestrogen also restored DLPFC functional capability in surgically induced menopausal rats (Hara). This increase was linked to positive changes in the structure of the presynaptic area (frequency of boutons and mitochondrial number).
We also ask whether factors such as emotions and age which are known influences on working memory performance can be associated with changes in DLPFC functioning. In the case of emotions, it is known that emotions, working memory task demands and individual cognitive differences predict behavior and cognitive effort. Negative emotional effects promote cognitive tendencies that are goal incompatible with task demands so that greater cognitive effort is required to perform well (Stobeck). This is supported by the observations that there is a negative impact of anxiety on working memory functioning (Chuderski) whereas positive feelings facilitate working memory and complex decision making among older adults (Carpenter). In this case, the effects of emotions could be explained by the close relationship between attention and working memory. It is known that changes to attentional awareness will elicit changes in informational quantity and quality of neural representations. From a neurochemical perspective, this translates to changes in firing and connectivity of appropriate brain areas and it has already been demonstrated above that the brain area important for personal values, the OFC, and the area discussed here, the DLPFC, demonstrate strong reciprocal connectivity. Therefore, any effects on OFC behaviour will influence DLPFC activity and it is possible that these changes can be affected through the NMDARs and AMPARs populations.
The other factor known to affect working memory performance is ageing. It is has been shown that the Gabergic dysfunction of the PFC in general with age leads to decreased working memory performance (Banuals). More specifically, we have already described that age-related cognitive decline is linked to the accessibility of NMDARs of the DLPFC of the particular sub-type that have the NR2A subunit. Therefore, if the activity of the DLPFC area is dependent on particular sub-populations of NMDARs then ageing may play a role in any performance change shown.
Therefore, we can conclude that NMDARs and AMPARs of the DLPFC play slightly differing roles during the transient world of working memory. The type of task reported on here required a number of different cognitive and sensory systems, eg. visual system, motor systems – head and eye movements, attentional system, working memory – information processing, short-term memory – visual memory, spatial memory and motor memory based and therefore, there are ample areas where NMDARs and AMPARs in the DLPFC could play a role. Van Vugt showed that AMPARs and NMDARs of the DLPFC area contributed to the persistent neuronal activity required for successful working memory employment during their experimental tasks. From their results they were able to conclude that NMDARs were essential for the short-term memory formation of the relevant information whereas AMPARs were involved in both relevant and irrelevant information whether transient or sustained. This was accounted for in neurochemical terms by the formation of sensory stores relating to the visual cue and then maintenance of these firing neurons in a short-term memory store representing their neuronal firing patterns. This is held in the working memory state whilst the distracting task of alternate visual cue is presented and extinguished. It was therefore, suggested that there was a subset of the firing population which was activated by the stimulus, but switched off when the stimulus was no longer visible and a set of ´delay cells` that exhibited persistent activity. And it is possible that these, as suggested by Van Vugt, are the sub-population of NMDARs that are linked to potassium channels, that of the SK channel which respond on binding of glutamate with an influx of potassium ions. It is assumed that if this is the case, then firing would be abolished and recovery of the cell instigated. Therefore, these neurons would represent those neurons responding to the presence of the actual visual cue whereas the other firing subset (ion channel type not determined) would maintain the neural representation in the working memory state (essentially the short-term memory). The contribution of AMPARs of the DLPFC in their working memory task were to the neural representation of the visual cues independent of location since they were shown to play a role in persistent activity. The functioning of the NMDARs and AMPARs of the DLPFC cells in response to visual stimuli in addition to the connectivity of the area would lead to the overall performance of working memory and other cognitive capabilities, such as executive control, the maintenance of items and value assessment. Therefore, studies on it are important and may provide subtle ways in which cognitive capability can be influenced.
Since we`re talking about the topic ………
….training has been shown to increase the strength of neuronal connectivity and working memory performance (Astle). Therefore, if subjects undergo a period of cognitive training before carrying out the ODR task can we assume that increases in firing potential and possibly differences to NMDAR and AMPAR contributions may be observed if Van Vugt`s experiments are repeated?
…Van Vugt`s experiments were performed without the monkeys being subjected to interference or distracting elements. If these were introduced and Van Vugt`s experiments were repeated, would we see as expected a decrease in performance accuracy due to working memory impairment, but no change in the relative NMDAR and AMPAR populations contributions in those subjects that performed the given tasks successfully?
…lesions of the basal ganglia lead to working memory being susceptible to irrelevant information (Baier). If these lesions were carried out and Van Vugt`s experiments repeated would we see the number of experimental failures increase as the monkeys were unable to carry out the demanded eye movements back to the target location due to the new visual stimulus location being given higher priority?
