iBet uBet web content aggregator. Adding the entire web to your favor.
iBet uBet web content aggregator. Adding the entire web to your favor.



Link to original content: http://www.ncbi.nlm.nih.gov/pubmed/32788365
Neuronal spike-rate adaptation supports working memory in language processing - PubMed Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2020 Aug 25;117(34):20881-20889.
doi: 10.1073/pnas.2000222117. Epub 2020 Aug 11.

Neuronal spike-rate adaptation supports working memory in language processing

Hartmut Fitz  1   2 , Marvin Uhlmann  2 , Dick van den Broek  2 , Renato Duarte  3   4   5 , Peter Hagoort  1   2 , Karl Magnus Petersson  6   2   7
Affiliations

Neuronal spike-rate adaptation supports working memory in language processing

Hartmut Fitz et al. Proc Natl Acad Sci U S A. .

Abstract

Language processing involves the ability to store and integrate pieces of information in working memory over short periods of time. According to the dominant view, information is maintained through sustained, elevated neural activity. Other work has argued that short-term synaptic facilitation can serve as a substrate of memory. Here we propose an account where memory is supported by intrinsic plasticity that downregulates neuronal firing rates. Single neuron responses are dependent on experience, and we show through simulations that these adaptive changes in excitability provide memory on timescales ranging from milliseconds to seconds. On this account, spiking activity writes information into coupled dynamic variables that control adaptation and move at slower timescales than the membrane potential. From these variables, information is continuously read back into the active membrane state for processing. This neuronal memory mechanism does not rely on persistent activity, excitatory feedback, or synaptic plasticity for storage. Instead, information is maintained in adaptive conductances that reduce firing rates and can be accessed directly without cued retrieval. Memory span is systematically related to both the time constant of adaptation and baseline levels of neuronal excitability. Interference effects within memory arise when adaptation is long lasting. We demonstrate that this mechanism is sensitive to context and serial order which makes it suitable for temporal integration in sequence processing within the language domain. We also show that it enables the binding of linguistic features over time within dynamic memory registers. This work provides a step toward a computational neurobiology of language.

Keywords: neuronal plasticity; sequence processing; working memory.

PubMed Disclaimer

Conflict of interest statement

The authors declare no competing interest.

Figures

Fig. 1.
Fig. 1.
Neuronal adaptation as a neurobiological correlate of WM. (A) Adaptive neuron is driven by a step current with amplitude 400 pA for a duration of 300 ms. Tonic spiking with uniformly spaced interspike intervals (Left) due to a refractory conductance gref (blue). Spike rate conductance gsra (red) adaptively decreases excitability and stretches out spike times (Right). Both conductances are spike-triggered but differ in magnitude and their decay time constants. (B) f–I curves show neuronal spike rates as a function of input current strength. As the time constant τsra of gsra increases (from 10 ms to 1.5 s; Left), spike rates decrease (black to gray gradient) because neuronal adaptation lasts longer. Likewise, as the magnitude of the spike-triggered change Δgsra increases (from 1 to 200 nS; Right), spike rates decrease (black to gray gradient) because adaptation becomes stronger. (C) Single neuron spike response (green dots) to a Poissonian input stimulus (orange; 0.5 kHz) from 50 presynaptic neurons, when preceded only by background noise (black; 0.25 kHz) or another sensory stimulus (blue; 0.5 kHz). (D) Histograms display SRA (dashed line indicates mean) which encodes context-dependent neuronal behavior in response to the orange stimulus. Memory of the blue stimulus is maintained in the hyperpolarized membrane state of the postsynaptic neuron (green). (E) Population-averaged memory traces gsra (red) over time, induced by a sequence of three items (y position of traces aligned with corresponding population). (F) Linear combination of these traces can distinguish the sequential order of inputs (123, 231, or 312) after stimulus offset (dashed line).
Fig. 2.
Fig. 2.
Network sequence processing. (A) Model comparison on semantic role assignment task in sentence comprehension. Accuracy is measured on all words in a sequence and on the sentence–final noun phrase. Spiking network outperforms memoryless logistic regression and perfect memory model which has access to the entire sentence context in WM. (B) Network accuracy improves with increasing time constant for neuronal adaptation. Peak performance occurs around τsra = 400 ms. Subsequent decline is due to interference in WM and can be prevented by flushing memory at the end of each sentence. Shaded regions indicate the size of interference effects on both measures. (C) Slow versus fast adapting neurons, controlled by the magnitude of the spike-triggered increase in adaptation conductance Δgsra. (D) Semantic role assignment accuracy parametrically varies with the degree of neuronal excitability for Δgsra ranging from 4 nS (slow adapting) to 500 nS (fast adapting). (E) Spiking network statistics (see SI Appendix for details): distribution of readout weights (log-scale), histograms of neuronal spike rates, coefficient of variation of interspike intervals (CV ISI), and pairwise spike synchrony (from top to bottom). Error bars in A, B, and D show 95% confidence intervals for 10 model subjects.
Fig. 3.
Fig. 3.
Binding of words to semantic roles. (A) After each input sentence, the network is queried with a semantic role label. The readout maps the network state onto a probability distribution of word responses for the queried role. A correct response occurs if the noun is identified that fills the query slot. (B) Feature binding accuracy for sentences with two occurrences of the target word as a function of the amount of language input. One readout identifies the lexical target, and the other readout returns the ordinal position of the target word. Error bars show 95% confidence intervals. (C) Example sentence and its trajectory through state space. Multiple occurrences of the same lexical noun (boy) in different semantic roles (agent, recipient) are separated by history-dependent neuronal processing.
Fig. 4.
Fig. 4.
Neurobiological read–write memory on multiple timescales. Sustained neural spiking activity has been viewed as a correlate of memory on short timescales. However, physiological processes other than the evolving membrane state provide dynamic variables for information storage on successively longer timescales. These include intrinsic plasticity, temporally extended synaptic currents, and short-term synaptic plasticity. Coupling of these processes to the membrane state creates read–write cycles where past information, held in slower dynamic variables (storage), is continuously folded back into the fast-changing, active network state (computation). The functional distinction between memory and computation is based on the timescales of dynamic variables.
See this image and copyright information in PMC

Similar articles

See all similar articles

Cited by

See all "Cited by" articles

References

    1. Goldman-Rakic P. S., Cellular basis of working memory. Neuron 14, 477–485 (1995). - PubMed
    1. Fuster J. M., Network memory. Trends Neurosci. 20, 451–459 (1997). - PubMed
    1. Wang X. J., Synaptic reverberation underlying mnemonic persistent activity. Trends Neurosci. 24, 455–463 (2001). - PubMed
    1. Amit D. J., Mongillo G., Selective delay activity in the cortex: Phenomena and interpretation. Cerebr. Cortex 13, 1139–1150 (2003). - PubMed
    1. Loewenstein Y., Sompolinsky H., Temporal integration by calcium dynamics in a model neuron. Nat. Neurosci. 6, 961–967 (2003). - PubMed

Publication types

LinkOut - more resources