Nice FMRI study showing that working memory delay activity is primarily in the superficial, feedforward, cortical layers while behavioral response-related activity is primarily in deep, feedback layers.
Layer-dependent activity in human prefrontal cortex during working memory
Emily S. Finn, Laurentius Huber, David C. Jangraw, Peter A. Bandettini
doi: https://doi.org/10.1101/425249
This is very consistent with our recent work:
Bastos, A.M., Loonis, R., Kornblith, S., Lundqvist, M., and Miller, E.K. (2018) Laminar recordings in frontal cortex suggest distinct layers for maintenance and control of working memory. Proceedings of the National Academy of Sciences. View PDF
Bouchacourt and Buschman describe a two-layer model of working memory. A sensory layer feeds into an unstructured layer of neurons with random connections (i.e., “mixed-selectivity” type neurons). It is flexible but interference between representations results in a capacity limit. Sounds like working memory to me.
Bouchacourt, F., & Buschman, T. J. (2018). A Flexible Model of Working Memory. bioRxiv, 407700.
More about mixed-selectivity:
Fusi, S., Miller, E.K., and Rigotti, M. (2016) Why neurons mix: High dimensionality for higher cognition. Current Opinion in Neurobiology. 37:66-74 doi:10.1016/j.conb.2016.01.010. View PDF »
Rigotti, M., Barak, O., Warden, M.R., Wang, X., Daw, N.D., Miller, E.K., & Fusi, S. (2013) The importance of mixed selectivity in complex cognitive tasks. Nature, 497, 585-590, doi:10.1038/nature12160. View PDF »
Nice summary of phase coding models of working memory by Hakim and Vogel, including a recent paper by Bahramisharif et al.
Hakim, N., & Vogel, E. K. (2018). Phase-coding memories in mind. PLoS biology, 16(8), e3000012.
Bahramisharif, A., Jensen, O., Jacobs, J., & Lisman, J. (2018). Serial representation of items during working memory maintenance at letter-selective cortical sites. PLoS biology, 16(8), e2003805.
Super-cool paper by Andreas Nieder and crew. Frontal-parietal beta synchrony encodes the most recent numerical input. Theta synchrony distinguishes between different numerosities held in working memory. The spiking of mixed-selectivity neurons multiplexed both task-relevant and irrelevant stimuli but they were separated in different phases of theta oscillations. Powerful support that neural oscillations functionally organize spiking activty.
Jacob, S. N., Hähnke, D., & Nieder, A. (2018). Structuring of Abstract Working Memory Content by Fronto-parietal Synchrony in Primate Cortex. Neuron, 99(3), 588-597.
Persistent activity (indexed by broadband gamma) across human cortex encodes stimulus features and predicts motor output.
Haller, Matar, John Case, Nathan E. Crone, Edward F. Chang, David King-Stephens, Kenneth D. Laxer, Peter B. Weber, Josef Parvizi, Robert T. Knight, and Avgusta Y. Shestyuk. “Persistent neuronal activity in human prefrontal cortex links perception and action.” Nature Human Behaviour (2017): 1.
But how persistent is it?
Lundqvist, M., Rose, J., Herman, P, Brincat, S.L, Buschman, T.J., and Miller, E.K. (2016) Gamma and beta bursts underlie working memory. Neuron, published online March 17, 2016. View PDF »
Gamma and beta bursts during working memory readout suggest roles in its volitional control
Lundqvist et al Nature Communications, in press.
Laminar recordings in frontal cortex suggest distinct layers for maintenance and control of working memory
Bastos et al PNAS, in press
Different levels of category abstraction by different dynamics in different prefrontal areas
Wutz et al Neuron, in press
Stay tuned for what they are about and what they mean. They add up to a new model of working memory.
Working memory in crows. Many of the same neural properties as primates.
Hartmann, K., Veit, L., & Nieder, A. (2017). Neurons in the crow nidopallium caudolaterale encode varying durations of visual working memory periods. Experimental Brain Research, 1-12.
Working memory for different items in a sequence is prioritized by how much attention is paid to the item at encoding.
Jafarpour, A., Penny, W., Barnes, G., Knight, R. T., & Duzel, E. (2017). Working Memory Replay Prioritizes Weakly Attended Events. eNeuro, 4(4), ENEURO-0171.
New result on bioRxiv:
Gamma and beta bursts during working memory read-out suggest roles in its volitional control
Mikael Lundqvist, Pawel Herman, Melissa R Warden, Scott L Brincat, Earl K Miller
doi: https://doi.org/10.1101/122598
Abstract
Working memory (WM) activity is not as stationary or sustained as previously thought. There are brief bursts of gamma (55 to 120 Hz) and beta (20 to 35 Hz) oscillations, the former linked to stimulus information in spiking. We examine these dynamics in relation to read-out from WM, which is still not well understood. Monkeys held a sequence of two objects and had to decide if they matched a subsequent sequence. Changes in the balance of beta/gamma suggested their role in WM control. In anticipation of having to use an object for the match decision, there was an increase in spiking information about that object along with an increase in gamma and a decrease in beta. When an object was no longer needed, beta increased and gamma as well as spiking information about that object decreased. Deviations from these dynamics predicted behavioral errors. Thus, turning up or down beta could regulate gamma and the information in working memory.
This review of the neural basis of working memory argues that working memory is a property of many brain areas working in concert. Prefrontal vs sensory cortical areas differ in their degrees of abstraction and how they are tied to action. They argue that the persistent activity that seems to underlie working memory is a general product of cortical networks.
Christophel, T. B., Klink, P. C., Spitzer, B., Roelfsema, P. R., & Haynes, J. D. (2017). The Distributed Nature of Working Memory. Trends in Cognitive Sciences.
I would add that persistent activity may not be so persistent:
Lunqvist, M., Rose, J., Herman, P, Brincat, S.L, Buschman, T.J., and Miller, E.K. (2016) Gamma and beta bursts underlie working memory. Neuron, published online March 17, 2016. View PDF »
Stokes, M., & Spaak, E. (2016). The Importance of Single-Trial Analyses in Cognitive Neuroscience. Trends in cognitive sciences.
Stokes, M. G. (2015). ‘Activity-silent’working memory in prefrontal cortex: a dynamic coding framework. Trends in cognitive sciences, 19(7), 394-405.
Stokes, M., Buschman, T.J., and Miller, E.K. (in press) Dynamic coding for flexible cognitive control. Wiley Handbook of Cognitive Control.
Ott and Nieder show that stimulating dopamine D2 receptors enhancing working memory related activity in the prefrontal cortex.
Ott, Torben, and Andreas Nieder. “Dopamine D2 Receptors Enhance Population Dynamics in Primate Prefrontal Working Memory Circuits.”Cerebral Cortex (2016).
Nice paper showing that different task demands in different task stages engage different oscillatory bands in the prefrontal cortex.
Wimmer, Klaus, et al. “Transitions between Multiband Oscillatory Patterns Characterize Memory-Guided Perceptual Decisions in Prefrontal Circuits.”The Journal of Neuroscience 36.2 (2016): 489-505.
Sustained activity has long been thought to be the neural substrate of working memory. But the evidence is based on averaging neural activity across trials. A closer examination reveals that something more complex is happening and supports a very different model of working memory.
Gamma and Beta Bursts Underlie Working Memory
Mikael Lundqvist, Jonas Rose, Pawel Herman, Scott L. Brincat, Timothy J. Buschman, Earl K. Miller
Neuron, published online March 17, 2016
Summary
Working memory is thought to result from sustained neuron spiking. However, computational models suggest complex dynamics with discrete oscillatory bursts. We analyzed local field potential (LFP) and spiking from the prefrontal cortex (PFC) of monkeys performing a working memory task. There were brief bursts of narrow-band gamma oscillations (45–100 Hz), varied in time and frequency, accompanying encoding and re-activation of sensory information. They appeared at a minority of recording sites associated with spiking reflecting the to-be-remembered items. Beta oscillations (20–35 Hz) also occurred in brief, variable bursts but reflected a default state interrupted by encoding and decoding. Only activity of neurons reflecting encoding/decoding correlated with changes in gamma burst rate. Thus, gamma bursts could gate access to, and prevent sensory interference with, working memory. This supports the hypothesis that working memory is manifested by discrete oscillatory dynamics and spiking, not sustained activity.
Nice review of our putative neural correlate of one of the most studied cognitive functions: Working memory.
Riley, Mitchell R., and Christos Constantinidis. “Role of Prefrontal Persistent Activity in Working Memory.” Frontiers in systems neuroscience 9 (2015).
Eriksson et al discuss working memory, not as an isolated function, but as an interaction between component processes such as attention, propsection, perception and long-term memory.
Eriksson, Johan, et al. “Neurocognitive Architecture of Working Memory.”Neuron 88.1 (2015): 33-46.
Ester et al use human imaging to show that the parietal and frontal cortices maintain information about specific visual stimuli held in memory. This shows that top-down control of working memory and storage functions are not so separate. We kind of knew that from the neuron level, but very nice demo in humans.
Ester, Edward F., Thomas C. Sprague, and John T. Serences. “Parietal and Frontal Cortex Encode Stimulus-Specific Mnemonic Representations during Visual Working Memory.” Neuron (2015).
Advance copy of our new paper:
Kornblith, S. Buschman, T.J., and Miller, E.K. (2015) Stimulus Load and Oscillatory Activity in Higher Cortex. Cerebral Cortex. doi: 10.1093/cercor/bhv182 Journal link
Abstract:
Exploring and exploiting a rich visual environment requires perceiving, attending, and remembering multiple objects simultaneously. Recent studies have suggested that this mental “juggling” of multiple objects may depend on oscillatory neural dynamics. We recorded local field potentials from the lateral intraparietal area, frontal eye fields, and lateral prefrontal cortex while monkeys maintained variable numbers of visual stimuli in working memory. Behavior suggested independent processing of stimuli in each hemifield. During stimulus presentation, higher-frequency power (50–100 Hz) increased with the number of stimuli (load) in the contralateral hemifield, whereas lower-frequency power (8–50 Hz) decreased with the total number of stimuli in both hemifields. During the memory delay, lower-frequency power increased with contralateral load. Load effects on higher frequencies during stimulus encoding and lower frequencies during the memory delay were stronger when neural activity also signaled the location of the stimuli. Like power, higher-frequency synchrony increased with load, but beta synchrony (16–30 Hz) showed the opposite effect, increasing when power decreased (stimulus presentation) and decreasing when power increased (memory delay). Our results suggest roles for lower-frequency oscillations in top-down processing and higher-frequency oscillations in bottom-up processing.
The limited capacity of working memory has sometimes been explained as a limited number of memory “slots”. Paul Bays argues that working memory capacity is due to sharing of a continuous resource, namely a fixed amount of neural activity. Noise in this activity is the limiting factor.
Working memory has long been thought to depend on sustained firing of cortical neurons. However, single neurons showing unbroken sustained activity is rare and average population activity is often only strong near the end of a memory delay. Mark Stokes presents the intriguing hypothesis for activity-silent working memory. He suggests that working memory depends on patterns of functional connectivity between neurons, not sustained activity.
‘Activity-silent’working memory in prefrontal cortex: a dynamic coding framework
MG Stokes – Trends in Cognitive Sciences, 2015
Kundu et al recorded EEG from humans during a short-term memory task. They found fronto-parietal coherence in different frequencies were associated with different memory functions. Alpha coherence was associated with maintenance of the information in memory. By contrast, the top-down filtering of distractions was associated with beta coherence. This adds to mounting evidence that specific frequency bands are associated with specific types of cortical processing like, for example, beta and top-down control.
Dotson et al recorded neural activity in the prefrontal and parietal cortex during a working memory task. As previous studies have reported (e.g., Buschman and Miller, 2007) they found long range synchronization of 8-25 Hz oscillations between the areas. Interestingly, there found both phase synchronization at 0 and 180 degrees suggesting that the 0 deg phase synchrony helped form networks between the areas whereas the 180 deg (anti-phase) synchrony helped segregate different networks.
For further reading:
Buschman, T.J. and Miller, E.K. (2007) Top-down versus bottom-up control of attention in the prefrontal and posterior parietal cortices. Science. 315: 1860-1862 View PDF »
Dotson et al report both 0 and 180 deg phase synchrony between the prefrontal and parietal cortices during a working memory task, suggestion both formation and segregation of different functional networks by neural synchrony.
A well-known correlate of working memory is sustained neural activity that bridges short gaps in time. It is well-established in the primate brain, but what about birds? They have working memory. (In fact, there is a lot of classic work that detailed the behavioral characteristics of working memory in pigeons).
Miller Lab alumnus Andreas Nieder and crew trained crows to perform a working memory task and found sustained activity in the nidopallium caudolaterale (NCL). This is presumably a neural correlate of the crow’s visual working memory.
Now if crows could only pass that causality test.
Anderson et al used scalp EEG recordings to decode the content of working memory and its quality. Subjects performed a orientation working memory task. Anderson et al found that the spatial distribution of alpha band power could be used to determine what orientation the subject was remembering and how precisely they were remembering it. Cool.
Matsushima and Tanaka compared neural correlates of spatial working memory for locations within the same hemifield or across hemifields. When the two remembered locations were in the same hemifield (right or left side of vision), the neural response in the prefrontal cortex was intermediate to the two cues presented alone. When the cues were across hemifields, the neural response was the same as the preferred cue presented alone. In other words, remembered locations within a hemifield seemed to be in competition with each other whereas locations across the hemifields seemed to be have no interaction at all. In yet other words, it was as if the (intact) monkeys had their brains split down the middle. The authors concluded local inhibitory interactions between cues within, but not across, hemifields.
This confirms Buschman et al (2011) who found that independent capacities for visual working memory in the right and left hemifields.
Further reading:
Buschman,T.J., Siegel, M., Roy, J.E. and Miller, E.K. (2011) Neural substrates of cognitive capacity limitations. Proceedings of the National Academy of Sciences. 108(27):11252-5. View PDF »