- Email: [email protected]
Neuroscience: A Mechanism for Rhythmic Sampling in Vision
Current Biology
Dispatches where their inputs and outputs are located will be critical to confirming what anatomical region they belong to, and how they fit into a broader circuit for processing social information. Another key question is what facet of the behavior the neurons are involved in. Could they, for example, be responsible for extracting social cues from visual stimuli, or for controlling the fishes’ interest in social interaction? Interestingly, neurons in the septum of chicks have been shown to respond to the appearance of conspecifics only when they move in a naturalistic way [19], suggesting a role in visual processing of social cues. Social attraction is observed already during the first weeks of life in zebrafish, at time points where the brain is still accessible for non-invasive calcium imaging approaches, and it will be exciting to discover if the current results can be extended to these earlier stages [13,14]. A proposed feature of the social behavior network is that different social contexts and behaviors are not processed by separate substrates, but are associated with distinct patterns of activation distributed across all the nodes in the network [1], and gene expression analysis in zebrafish has provided support for this idea [20]. It will therefore be interesting to know if ablation of these neurons affects the expression of other social behaviors such as aggression and courtship. The work of Stednitz et al. [3] makes it possible to begin addressing these questions, and to dissect the neural circuit mechanisms of social behavior in a simple vertebrate model.
light-sheet microscopy. Nat. Methods 10, 413–420. 6. Portugues, R., Feierstein, C.E., Engert, F., and Orger, M.B. (2014). Whole-brain activity maps reveal stereotyped, distributed networks for visuomotor behavior. Neuron 81, 1328–1343. 7. Kim, D.H., Kim, J., Marques, J.C., Grama, A., Hildebrand, D.G.C., Gu, W., Li, J.M., and Robson, D.N. (2017). Pan-neuronal calcium imaging with cellular resolution in freely swimming zebra fish. Nat. Methods 14, 1107–1114. 8. Orger, M.B., and de Polavieja, G.G. (2017). Zebrafish behavior: opportunities and challenges. Annu. Rev. Neurosci. 40, 125–147. 9. Abril-de Abreu, R., Cruz, J., and Oliveira, J.F. (2015). Social eavesdropping in zebrafish: tuning of attention to social interactions. Sci. Rep. 5, 11–14. 10. Engeszer, R.E., Ryan, M.J., and Parichy, D.M. (2014). Learned social preference in zebrafish. Curr. Biol. 14, 881–884. 11. Neri, P. (2012). Feature binding in zebrafish. Anim. Behav. 84, 485–493. 12. Chou, M.-Y., Amo, R., Kinoshita, M., Cherng, B.-W., Shimazaki, H., Agetsuma, M., Shiraki, T., Aoki, T., Takahoko, M., Yamazaki, M., et al. (2016). Social conflict resolution regulated by two dorsal habenular subregions in zebrafish. Science 352, 87–90. 13. Dreosti, E., Lopes, G., Kamp, A.R., and Wilson, S.W. (2015). Development of social behavior in young zebrafish. Front. Neural Circuits 9, 1–9.
14. Hinz, R.C., and de Polavieja, G.G. (2017). Ontogeny of collective behavior reveals a simple attraction rule. Proc. Natl. Acad. Sci. USA 114, 2295–2300. 15. Butail, S., Polverino, G., Phamduy, P., Del Sette, F., and Porfiri, M. (2014). Influence of robotic shoal size, configuration, and activity on zebrafish behavior in a free-swimming environment. Behav. Brain Res. 275, 269–280. 16. Stowers, J.R., Hofbauer, M., Bastien, R., Griessner, J., Higgins, P., Farooqui, S., Fischer, R.M., Nowikovsky, K., Haubensak, W., Couzin, I.D., et al. (2017). Virtual reality for freely moving animals. Nat. Methods 14, 995–1002. 17. Shinozuka, K., and Watanabe, S. (2004). Effects of telencephalic ablation on shoaling behavior in goldfish. Phys. Behav. 81, 141–148. 18. Marquart, G.D., Tabor, K.M., Brown, M., Strykowski, J.L., Varshney, G.K., LaFave, M.C., Mueller, T., Burgess, S.M., Higashijima, S., and Burgess, H.A. (2015). A 3D searchable database of transgenic zebrafish Gal4 and Cre lines for functional neuroanatomy studies. Front. Neural Circuits 9, 11. 19. Mayer, U., Rosa-Salva, O., Morbioli, F., and Vallortigara, G. (2017). The motion of a living conspecific activates septal and preoptic areas in naive domestic chicks (Gallus gallus). Eur. J. Neurosci. 45, 423–432. 20. Teles, M.C., Cardoso, S.D., and Oliveira, R.F. (2016). Social plasticity relies on different neuroplasticity mechanisms across the brain social decision-making network in zebrafish. Front. Behav. Neurosci. 10, 16.
Neuroscience: A Mechanism for Rhythmic Sampling in Vision Ayelet N. Landau
REFERENCES 1. Newman, S.W. (1999). The medial extended amygdala in male reproductive behavior. A node in the mammalian social behavior network. Ann. N.Y. Acad. Sci. 877, 242–257. 2. O’Connell, L.A., and Hofmann, H.A. (2012). Evolution of a vertebrate social decisionmaking network. Science 336, 1154–1157. 3. Stednitz, S.J., McDermott, E.M., Ncube, D., Tallafuss, A., Eisen, J.S., and Washbourne, P. (2018). Forebrain control of behaviorally-driven social orienting in zebrafish. Curr. Biol. 28, 2445–2451. 4. Kawakami, K., Asakawa, K., Hibi, M., Itoh, M., Muto, A., and Wada, H. (2016). Gal4 driver transgenic zebrafish. Adv. Gen. 95, 65–87. 5. Ahrens, M.B., Orger, M.B., Robson, D.N., Li, J.M., and Keller, P.J. (2013). Whole-brain functional imaging at cellular resolution using
Departments of Psychology and Cognitive Sciences, The Hebrew University of Jerusalem, Jerusalem, Israel Correspondence: [email protected] https://doi.org/10.1016/j.cub.2018.05.081
Ongoing perception ebbs and flows rhythmically. Understanding the source and scope of this phenomenon is an important step in unraveling the foundations of sensory processing. A new study demonstrates that local neuronal interactions generate rhythmic brain activity and correspond to rhythmic performance patterns on a visual-detection task. Researchers in the fields of psychology and neuroscience often debate different taxonomies and definitions for attention [1]. In physiology, the act-by-act consequences of attention can readily be described in the visual cortex and revolve
R830 Current Biology 28, R826–R845, August 6, 2018 ª 2018 Elsevier Ltd.
around two key concepts: competition and bias [2]. When neurons in the visual cortex encounter a cluttered scene, the different stimuli compete for the sensory neurons’ response. If we were to ‘listen’ to (measure) a visually responsive neuron, it
Current Biology
Dispatches would be very difficult to know what it is ‘seeing’ (responding to) when several stimuli fall in its receptive field (the part of the visual field it responds to). This competitive interaction between multiple stimuli can be resolved if attention were directed to one of the items in the scene. Attention serves as a biasing signal; when one item in the clutter is attended, a particular neuron will respond as if the item were presented in isolation. If we were to ‘listen’ to the neuronal response now, we could delineate what part of the cluttered scene it is ‘seeing’. In fact, we might mistakenly conclude that the attended item is the only one presented! This proves that attention causes neurons to tune in to stimuli that are relevant, over other, irrelevant stimuli; this phenomenon is termed biased competition. Biased competition describes how attention operates in many different scenarios [2,3]. Most of these scenarios, however, entail attending to a single relevant object or location. In the natural environment, we face a constant stream of sensory input, and often more than one location is relevant. How does attention unfold over multiple locations? Attentional sampling, measured initially in behavioral performance [4], provides a good starting point to answer this question: When two locations are relevant we alternate between them, sequentially sampling each location. Each second entails eight samples. If the display includes two relevant locations, those eight samples will be distributed, four per location, in alternation (theta rhythm) [4]. The basic finding of rhythmic sampling in alternation has been replicated and extended in recent years [5–10]. As they report in this issue of Current Biology, Kienitz et al. [11] investigated a neural mechanism of attentional sampling. They were able to delineate the detailed interplay between nearby neuronal populations that generates theta-rhythmic fluctuation (4–8 Hz), and which corresponds to a theta rhythm in behavioral performance. The proposed mechanism relates to basic principles in the response properties of neurons in visual cortex. Many visual neurons have the following response profile (Figure 1C): when a stimulus is presented to the central part of a neuron’s receptive field, this generates an excitatory response; but when a nearby stimulus is
A
B
C
8 Hz Rhythmic selection (control regions)
- --
-
+ +
-
--
-
8 Hz Local interaction (sensory substrate)
1 second
1 second
1 second Current Biology
Figure 1. Models for the emergence of theta rhythmic sampling. (A) A common sampler account in which attentional-control mechanisms implement theta rhythmic sampling in alternation. This model entails that the content of each sample is determined by a control region or network that is exerting the biasing attention signals onto sensory processors. Locally, within sensory areas a 4 Hz rhythm is measured, similar to that measured in performance. However, the sampler operates at double that frequency (8 Hz) and distributes the samples in alternation. Red and blue circles denote two items, which are attended. They can be within a system or between systems, for example those for different sensory modalities. The signals correspondingly mark the neural response from the relevant population representing each item. (B) Rhythmic sampling emerging from local interactions wherein the representation of one item (for example, red circle) generates the suppressive impact on the other item (blue circle). In this model, an 8 Hz rhythm governs excitability within sensory cortex (black signal) and the rhythmic alternation results from the relative impact one item has on the other. The thick vertical red and blue lines denote the item represented at each cycle. The possibility of a phase-code is included in this diagram whereby items are represented in a sequence during each cycle (at varying degrees, denoted by line width and saturation). This possibility is elaborated on in [19]. (C) When two items are nearby, mutual inhibition can arise from the architecture of classical receptive fields. In this diagram, adapted from Kienitz et al. [9], each item is centered on a neuronal receptive field (thin, concentric circles around each item). Each item (red and blue discs) occupies the central, excitatory portion of one respective receptive field, and the surrounding inhibitory portion of the neighboring receptive field. This organization leads to rhythmic neuronal activity reported in the new paper from Kienitz et al. [9]. The signals in this panel, therefore, follow the idealized receptive field response for each item. The rhythmic pattern of this response is the result of mutual inhibition of these receptive fields. Contrary to panel B, where the temporal structure emerges due to a generalized inhibitory drive (the alpha oscillation), in this scenario the temporal structure emerges in light of the nearby receptive field interaction.
presented — falling within the receptive field’s surround — the neuron’s response is attenuated. Kienitz et al. [11] found that theta rhythmic neural activity emerged in the response of nearby neurons when those were stimulated with adjacent stimuli (so that center and surround regions were stimulated, each by a different stimulus; Figure 1C). This finding raises the possibility that in vision, attentional sampling — the rhythmic profile of our perceptions — may, in fact, rely on the local architecture of visual neurons. In what follows I will contextualize the current findings within putative accounts for attentional sampling. Understanding Attentional Sampling One possible explanation for attentional sampling is that the biasing signal itself, in other words, attention itself, is a rhythmic process (Figure 1A). Attentional
control is achieved by the engagement of large-scale brain networks [1,12]. This architecture gives rise to the possibility that attentional sampling is, in fact, a property of these control networks or regions. When individuals perform a distributed attention task, they are requested to voluntarily monitor two locations in order to detect a target. Such instructions unavoidably engage attentional control systems in the brain. The discovery that this process results in rhythmic performance patterns could mean that the selection process itself dictates the rhythmic pattern of performance (Figure 1A). In other words, a central sampler within large-scale networks determines the content of each sample and accounts for the alternations in performance and the corresponding alternations in ongoing brain activity [13,14]. Notably, this account is domaingeneral and could provide a framework
Current Biology 28, R826–R845, August 6, 2018 R831
Current Biology
Dispatches for understanding how attention coordinates inputs from different brain systems (for example, those orchestrating different sensory modalities). Consistent with this domaingeneral property, attentional sampling has been characterized both in and between different systems, such as those for action, eye movements, and audition [7,8,15,16]. Alternatively, the content of a sample could be the result of local interactions between neuronal populations within the sensory system (Figure 1B,C). One observation that may support this idea is the ubiquitous measurement of neural oscillations. Electroencephalography (EEG) in humans revealed, almost a century ago [17], a robust oscillation: the visual alpha rhythm (8–10 Hz). Robust fluctuations in cortical excitability — the oscillation — inevitably shapes the temporal structure of sensory responses. A stimulus can be represented or sequentially explored according to the excitation it elicits with respect to the rhythmic cortical fluctuations (Figure 1B). In the case of two objects, post-excitation inhibition of a represented object, or a form of mutual inhibition between two objects (Figure 1C), could account for alternations from one object to the next — from one alpha cycle to the next (Figure 1B) ([18,19] and Ole Jensen, personal communication). Importantly, this type of model suggests that the content of each cycle is resolved and determined, not according to a networkwide process, but due to relative excitation between adjacent neurons in a visual area. Consistent with this view, Kienitz et al. [11] link attentional sampling to a form of mutual inhibition that results from receptive field properties of neighboring neurons (Figure 1C). They were able to show that, when two nearby stimuli are either passively viewed or both attended, then neuronal activity fluctuates in alternation in a pattern consistent with the behavioral performance. The authors did not have a condition that eliminates the engagement of attention in the task; however, the described local neuronal interactions, which generate the thetarhythmic fluctuations, are a known property of receptive fields that has been measured in anesthetized animals. In light of this, I suggest that attentional sampling
may be better termed rhythmic sampling, as the suggested mechanism appears orthogonal to the deployment of attention (see also [20]). The Future of Rhythmic Sampling The data presented by Kienitz et al. [11] are compelling, and capture the phenomenon of rhythmic sampling in behavior together with a convincing neural mechanism, basic to neural representation — visual receptive-field interactions. An important next step will be to clarify how these findings and proposed mechanisms relate to findings from behavior and non-invasive physiology. In particular, many studies report rhythmic sampling between two locations spaced across different hemifields [4,5,7,13]. Can adjacentreceptive field interactions generalize over such spatial scales? Moreover, could the mechanism described here account for rhythmic sampling between different systems, such as the visual and motor systems [8]? Finally, can the mechanism generalize to rhythmic sampling in the absence of nearby receptive field stimulation, for example, in the case of performance at a single location [14]? The link between the admirably detailed description by Kienitz et al. [11] and the rest of the rhythmic sampling literature, measured at a larger scale, lays the groundwork for an exciting path forward in delineating the architecture of ongoing perceptual processes. REFERENCES 1. Nobre, A.C. (2018). Attention. In Stevens’ Handbook of Experimental Psychology and Cognitive Neuroscience, Sensation, Perception, and Attention, J. Wixted, and J. Serences, eds. (New York: John Wiley & Sons, Inc.), p. 241. 2. Reynolds, J.H., Chelazzi, L., and Desimone, R. (1999). Competitive mechanisms subserve attention in macaque areas V2 and V4. J. Neurosci. 19, 1736–1753. 3. Beck, D.M., and Kastner, S. (2009). Top-down and bottom-up mechanisms in biasing competition in the human brain. Vision Res. 49, 1154–1165.
6. Song, K., Meng, M., Chen, L., Zhou, K., and Luo, H. (2014). Behavioral oscillations in Attention: rhythmic alpha pulses mediated through theta band. J. Neurosci. 34, 4837– 4844. 7. Hogendoorn, H. (2016). Voluntary saccadic eye movements ride the attentional rhythm. J. Cogn. Neurosci. 28, 1625–1635. 8. Tomassini, A., Spinelli, D., Jacono, M., Sandini, G., and Morrone, M.C. (2015). Rhythmic Oscillations of Visual Contrast Sensitivity Synchronized with Action. J. Neurosci. 35, 7019–7029. , L., McLelland, D., Lajous, M., and 9. Dugue VanRullen, R. (2015). Attention searches nonuniformly in space and in time. Proc. Natl. Acad. Sci. USA 112, 15214–15219. 10. Benedetto, A., Spinelli, D., Morrone, M.C., Spaak, E., Lange, F., de Jensen, O., Tomassini, A., Spinelli, D., Jacono, M., Sandini, G., et al. (2016). Rhythmic modulation of visual contrast discrimination triggered by action. Proc. Biol. Sci. 283, 3536–3544. 11. Kienitz, R., Schmiedt, J.T., Shapcott, K.A., Kouroupaki, K., Saunders, R.C., and Schmid, M.C. (2018). Theta rhythmic neuronal activity and reaction times arising from cortical receptive field interactions during distributed attention. Curr. Biol. 28, 2377– 2387. 12. Buschman, T.J., and Kastner, S. (2015). From behavior to neural dynamics: an integrated theory of attention. Neuron 88, 127–144. 13. Landau, A.N., Schreyer, H.M., van Pelt, S., and Fries, P. (2015). Distributed attention is implemented through theta-rhythmic gamma modulation. Curr. Biol. 25, 2332–2337. 14. VanRullen, R. (2016). Perceptual cycles. Trends Cogn. Sci. 20, 723–735. 15. Ho, H.T., Leung, J., Burr, D.C., Alais, D., and Morrone, M.C. (2017). Auditory sensitivity and decision criteria oscillate at different frequencies separately for the two ears. Curr. Biol. 27, 3643–3649.e3. 16. Tomassini, A., Ambrogioni, L., Medendorp, W.P., and Maris, E. (2017). Theta oscillations locked to intended actions rhythmically modulate perception. Elife 6, 1–18. 17. Berger, H. (1930). Uber das Elektrenkephalogramm des Menschen. Zweite Mitteilung. J. Psycho. Neurol. 40, 160–179. 18. de Almeida, L., Idiart, M., and Lisman, J.E. (2009). A second function of gamma frequency oscillations: an E%-Max winner-take-all mechanism selects which cells fire. J. Neurosci. 29, 7497–7503.
4. Landau, A.N., and Fries, P. (2012). Attention samples stimuli rhythmically. Curr. Biol. 22, 1000–1004.
19. Jensen, O., Gips, B., Bergmann, T.O., and Bonnefond, M. (2014). Temporal coding organized by coupled alpha and gamma oscillations prioritize visual processing. Trends Neurosci. 37, 357–369.
5. Fiebelkorn, I.C., Saalmann, Y.B., and Kastner, S. (2013). Rhythmic sampling within and between objects despite sustained attention at a cued location. Curr. Biol. 23, 2553–2558.
20. Spyropoulos, G., Bosman, C.A., and Fries, P. (2017). A theta rhythm in awake macaque V1 and V4 and its attentional modulation. bioRxiv, https://doi.org/10.1101/117804.
R832 Current Biology 28, R826–R845, August 6, 2018
Dispatches where their inputs and outputs are located will be critical to confirming what anatomical region they belong to, and how they fit into a broader circuit for processing social information. Another key question is what facet of the behavior the neurons are involved in. Could they, for example, be responsible for extracting social cues from visual stimuli, or for controlling the fishes’ interest in social interaction? Interestingly, neurons in the septum of chicks have been shown to respond to the appearance of conspecifics only when they move in a naturalistic way [19], suggesting a role in visual processing of social cues. Social attraction is observed already during the first weeks of life in zebrafish, at time points where the brain is still accessible for non-invasive calcium imaging approaches, and it will be exciting to discover if the current results can be extended to these earlier stages [13,14]. A proposed feature of the social behavior network is that different social contexts and behaviors are not processed by separate substrates, but are associated with distinct patterns of activation distributed across all the nodes in the network [1], and gene expression analysis in zebrafish has provided support for this idea [20]. It will therefore be interesting to know if ablation of these neurons affects the expression of other social behaviors such as aggression and courtship. The work of Stednitz et al. [3] makes it possible to begin addressing these questions, and to dissect the neural circuit mechanisms of social behavior in a simple vertebrate model.
light-sheet microscopy. Nat. Methods 10, 413–420. 6. Portugues, R., Feierstein, C.E., Engert, F., and Orger, M.B. (2014). Whole-brain activity maps reveal stereotyped, distributed networks for visuomotor behavior. Neuron 81, 1328–1343. 7. Kim, D.H., Kim, J., Marques, J.C., Grama, A., Hildebrand, D.G.C., Gu, W., Li, J.M., and Robson, D.N. (2017). Pan-neuronal calcium imaging with cellular resolution in freely swimming zebra fish. Nat. Methods 14, 1107–1114. 8. Orger, M.B., and de Polavieja, G.G. (2017). Zebrafish behavior: opportunities and challenges. Annu. Rev. Neurosci. 40, 125–147. 9. Abril-de Abreu, R., Cruz, J., and Oliveira, J.F. (2015). Social eavesdropping in zebrafish: tuning of attention to social interactions. Sci. Rep. 5, 11–14. 10. Engeszer, R.E., Ryan, M.J., and Parichy, D.M. (2014). Learned social preference in zebrafish. Curr. Biol. 14, 881–884. 11. Neri, P. (2012). Feature binding in zebrafish. Anim. Behav. 84, 485–493. 12. Chou, M.-Y., Amo, R., Kinoshita, M., Cherng, B.-W., Shimazaki, H., Agetsuma, M., Shiraki, T., Aoki, T., Takahoko, M., Yamazaki, M., et al. (2016). Social conflict resolution regulated by two dorsal habenular subregions in zebrafish. Science 352, 87–90. 13. Dreosti, E., Lopes, G., Kamp, A.R., and Wilson, S.W. (2015). Development of social behavior in young zebrafish. Front. Neural Circuits 9, 1–9.
14. Hinz, R.C., and de Polavieja, G.G. (2017). Ontogeny of collective behavior reveals a simple attraction rule. Proc. Natl. Acad. Sci. USA 114, 2295–2300. 15. Butail, S., Polverino, G., Phamduy, P., Del Sette, F., and Porfiri, M. (2014). Influence of robotic shoal size, configuration, and activity on zebrafish behavior in a free-swimming environment. Behav. Brain Res. 275, 269–280. 16. Stowers, J.R., Hofbauer, M., Bastien, R., Griessner, J., Higgins, P., Farooqui, S., Fischer, R.M., Nowikovsky, K., Haubensak, W., Couzin, I.D., et al. (2017). Virtual reality for freely moving animals. Nat. Methods 14, 995–1002. 17. Shinozuka, K., and Watanabe, S. (2004). Effects of telencephalic ablation on shoaling behavior in goldfish. Phys. Behav. 81, 141–148. 18. Marquart, G.D., Tabor, K.M., Brown, M., Strykowski, J.L., Varshney, G.K., LaFave, M.C., Mueller, T., Burgess, S.M., Higashijima, S., and Burgess, H.A. (2015). A 3D searchable database of transgenic zebrafish Gal4 and Cre lines for functional neuroanatomy studies. Front. Neural Circuits 9, 11. 19. Mayer, U., Rosa-Salva, O., Morbioli, F., and Vallortigara, G. (2017). The motion of a living conspecific activates septal and preoptic areas in naive domestic chicks (Gallus gallus). Eur. J. Neurosci. 45, 423–432. 20. Teles, M.C., Cardoso, S.D., and Oliveira, R.F. (2016). Social plasticity relies on different neuroplasticity mechanisms across the brain social decision-making network in zebrafish. Front. Behav. Neurosci. 10, 16.
Neuroscience: A Mechanism for Rhythmic Sampling in Vision Ayelet N. Landau
REFERENCES 1. Newman, S.W. (1999). The medial extended amygdala in male reproductive behavior. A node in the mammalian social behavior network. Ann. N.Y. Acad. Sci. 877, 242–257. 2. O’Connell, L.A., and Hofmann, H.A. (2012). Evolution of a vertebrate social decisionmaking network. Science 336, 1154–1157. 3. Stednitz, S.J., McDermott, E.M., Ncube, D., Tallafuss, A., Eisen, J.S., and Washbourne, P. (2018). Forebrain control of behaviorally-driven social orienting in zebrafish. Curr. Biol. 28, 2445–2451. 4. Kawakami, K., Asakawa, K., Hibi, M., Itoh, M., Muto, A., and Wada, H. (2016). Gal4 driver transgenic zebrafish. Adv. Gen. 95, 65–87. 5. Ahrens, M.B., Orger, M.B., Robson, D.N., Li, J.M., and Keller, P.J. (2013). Whole-brain functional imaging at cellular resolution using
Departments of Psychology and Cognitive Sciences, The Hebrew University of Jerusalem, Jerusalem, Israel Correspondence: [email protected] https://doi.org/10.1016/j.cub.2018.05.081
Ongoing perception ebbs and flows rhythmically. Understanding the source and scope of this phenomenon is an important step in unraveling the foundations of sensory processing. A new study demonstrates that local neuronal interactions generate rhythmic brain activity and correspond to rhythmic performance patterns on a visual-detection task. Researchers in the fields of psychology and neuroscience often debate different taxonomies and definitions for attention [1]. In physiology, the act-by-act consequences of attention can readily be described in the visual cortex and revolve
R830 Current Biology 28, R826–R845, August 6, 2018 ª 2018 Elsevier Ltd.
around two key concepts: competition and bias [2]. When neurons in the visual cortex encounter a cluttered scene, the different stimuli compete for the sensory neurons’ response. If we were to ‘listen’ to (measure) a visually responsive neuron, it
Current Biology
Dispatches would be very difficult to know what it is ‘seeing’ (responding to) when several stimuli fall in its receptive field (the part of the visual field it responds to). This competitive interaction between multiple stimuli can be resolved if attention were directed to one of the items in the scene. Attention serves as a biasing signal; when one item in the clutter is attended, a particular neuron will respond as if the item were presented in isolation. If we were to ‘listen’ to the neuronal response now, we could delineate what part of the cluttered scene it is ‘seeing’. In fact, we might mistakenly conclude that the attended item is the only one presented! This proves that attention causes neurons to tune in to stimuli that are relevant, over other, irrelevant stimuli; this phenomenon is termed biased competition. Biased competition describes how attention operates in many different scenarios [2,3]. Most of these scenarios, however, entail attending to a single relevant object or location. In the natural environment, we face a constant stream of sensory input, and often more than one location is relevant. How does attention unfold over multiple locations? Attentional sampling, measured initially in behavioral performance [4], provides a good starting point to answer this question: When two locations are relevant we alternate between them, sequentially sampling each location. Each second entails eight samples. If the display includes two relevant locations, those eight samples will be distributed, four per location, in alternation (theta rhythm) [4]. The basic finding of rhythmic sampling in alternation has been replicated and extended in recent years [5–10]. As they report in this issue of Current Biology, Kienitz et al. [11] investigated a neural mechanism of attentional sampling. They were able to delineate the detailed interplay between nearby neuronal populations that generates theta-rhythmic fluctuation (4–8 Hz), and which corresponds to a theta rhythm in behavioral performance. The proposed mechanism relates to basic principles in the response properties of neurons in visual cortex. Many visual neurons have the following response profile (Figure 1C): when a stimulus is presented to the central part of a neuron’s receptive field, this generates an excitatory response; but when a nearby stimulus is
A
B
C
8 Hz Rhythmic selection (control regions)
- --
-
+ +
-
--
-
8 Hz Local interaction (sensory substrate)
1 second
1 second
1 second Current Biology
Figure 1. Models for the emergence of theta rhythmic sampling. (A) A common sampler account in which attentional-control mechanisms implement theta rhythmic sampling in alternation. This model entails that the content of each sample is determined by a control region or network that is exerting the biasing attention signals onto sensory processors. Locally, within sensory areas a 4 Hz rhythm is measured, similar to that measured in performance. However, the sampler operates at double that frequency (8 Hz) and distributes the samples in alternation. Red and blue circles denote two items, which are attended. They can be within a system or between systems, for example those for different sensory modalities. The signals correspondingly mark the neural response from the relevant population representing each item. (B) Rhythmic sampling emerging from local interactions wherein the representation of one item (for example, red circle) generates the suppressive impact on the other item (blue circle). In this model, an 8 Hz rhythm governs excitability within sensory cortex (black signal) and the rhythmic alternation results from the relative impact one item has on the other. The thick vertical red and blue lines denote the item represented at each cycle. The possibility of a phase-code is included in this diagram whereby items are represented in a sequence during each cycle (at varying degrees, denoted by line width and saturation). This possibility is elaborated on in [19]. (C) When two items are nearby, mutual inhibition can arise from the architecture of classical receptive fields. In this diagram, adapted from Kienitz et al. [9], each item is centered on a neuronal receptive field (thin, concentric circles around each item). Each item (red and blue discs) occupies the central, excitatory portion of one respective receptive field, and the surrounding inhibitory portion of the neighboring receptive field. This organization leads to rhythmic neuronal activity reported in the new paper from Kienitz et al. [9]. The signals in this panel, therefore, follow the idealized receptive field response for each item. The rhythmic pattern of this response is the result of mutual inhibition of these receptive fields. Contrary to panel B, where the temporal structure emerges due to a generalized inhibitory drive (the alpha oscillation), in this scenario the temporal structure emerges in light of the nearby receptive field interaction.
presented — falling within the receptive field’s surround — the neuron’s response is attenuated. Kienitz et al. [11] found that theta rhythmic neural activity emerged in the response of nearby neurons when those were stimulated with adjacent stimuli (so that center and surround regions were stimulated, each by a different stimulus; Figure 1C). This finding raises the possibility that in vision, attentional sampling — the rhythmic profile of our perceptions — may, in fact, rely on the local architecture of visual neurons. In what follows I will contextualize the current findings within putative accounts for attentional sampling. Understanding Attentional Sampling One possible explanation for attentional sampling is that the biasing signal itself, in other words, attention itself, is a rhythmic process (Figure 1A). Attentional
control is achieved by the engagement of large-scale brain networks [1,12]. This architecture gives rise to the possibility that attentional sampling is, in fact, a property of these control networks or regions. When individuals perform a distributed attention task, they are requested to voluntarily monitor two locations in order to detect a target. Such instructions unavoidably engage attentional control systems in the brain. The discovery that this process results in rhythmic performance patterns could mean that the selection process itself dictates the rhythmic pattern of performance (Figure 1A). In other words, a central sampler within large-scale networks determines the content of each sample and accounts for the alternations in performance and the corresponding alternations in ongoing brain activity [13,14]. Notably, this account is domaingeneral and could provide a framework
Current Biology 28, R826–R845, August 6, 2018 R831
Current Biology
Dispatches for understanding how attention coordinates inputs from different brain systems (for example, those orchestrating different sensory modalities). Consistent with this domaingeneral property, attentional sampling has been characterized both in and between different systems, such as those for action, eye movements, and audition [7,8,15,16]. Alternatively, the content of a sample could be the result of local interactions between neuronal populations within the sensory system (Figure 1B,C). One observation that may support this idea is the ubiquitous measurement of neural oscillations. Electroencephalography (EEG) in humans revealed, almost a century ago [17], a robust oscillation: the visual alpha rhythm (8–10 Hz). Robust fluctuations in cortical excitability — the oscillation — inevitably shapes the temporal structure of sensory responses. A stimulus can be represented or sequentially explored according to the excitation it elicits with respect to the rhythmic cortical fluctuations (Figure 1B). In the case of two objects, post-excitation inhibition of a represented object, or a form of mutual inhibition between two objects (Figure 1C), could account for alternations from one object to the next — from one alpha cycle to the next (Figure 1B) ([18,19] and Ole Jensen, personal communication). Importantly, this type of model suggests that the content of each cycle is resolved and determined, not according to a networkwide process, but due to relative excitation between adjacent neurons in a visual area. Consistent with this view, Kienitz et al. [11] link attentional sampling to a form of mutual inhibition that results from receptive field properties of neighboring neurons (Figure 1C). They were able to show that, when two nearby stimuli are either passively viewed or both attended, then neuronal activity fluctuates in alternation in a pattern consistent with the behavioral performance. The authors did not have a condition that eliminates the engagement of attention in the task; however, the described local neuronal interactions, which generate the thetarhythmic fluctuations, are a known property of receptive fields that has been measured in anesthetized animals. In light of this, I suggest that attentional sampling
may be better termed rhythmic sampling, as the suggested mechanism appears orthogonal to the deployment of attention (see also [20]). The Future of Rhythmic Sampling The data presented by Kienitz et al. [11] are compelling, and capture the phenomenon of rhythmic sampling in behavior together with a convincing neural mechanism, basic to neural representation — visual receptive-field interactions. An important next step will be to clarify how these findings and proposed mechanisms relate to findings from behavior and non-invasive physiology. In particular, many studies report rhythmic sampling between two locations spaced across different hemifields [4,5,7,13]. Can adjacentreceptive field interactions generalize over such spatial scales? Moreover, could the mechanism described here account for rhythmic sampling between different systems, such as the visual and motor systems [8]? Finally, can the mechanism generalize to rhythmic sampling in the absence of nearby receptive field stimulation, for example, in the case of performance at a single location [14]? The link between the admirably detailed description by Kienitz et al. [11] and the rest of the rhythmic sampling literature, measured at a larger scale, lays the groundwork for an exciting path forward in delineating the architecture of ongoing perceptual processes. REFERENCES 1. Nobre, A.C. (2018). Attention. In Stevens’ Handbook of Experimental Psychology and Cognitive Neuroscience, Sensation, Perception, and Attention, J. Wixted, and J. Serences, eds. (New York: John Wiley & Sons, Inc.), p. 241. 2. Reynolds, J.H., Chelazzi, L., and Desimone, R. (1999). Competitive mechanisms subserve attention in macaque areas V2 and V4. J. Neurosci. 19, 1736–1753. 3. Beck, D.M., and Kastner, S. (2009). Top-down and bottom-up mechanisms in biasing competition in the human brain. Vision Res. 49, 1154–1165.
6. Song, K., Meng, M., Chen, L., Zhou, K., and Luo, H. (2014). Behavioral oscillations in Attention: rhythmic alpha pulses mediated through theta band. J. Neurosci. 34, 4837– 4844. 7. Hogendoorn, H. (2016). Voluntary saccadic eye movements ride the attentional rhythm. J. Cogn. Neurosci. 28, 1625–1635. 8. Tomassini, A., Spinelli, D., Jacono, M., Sandini, G., and Morrone, M.C. (2015). Rhythmic Oscillations of Visual Contrast Sensitivity Synchronized with Action. J. Neurosci. 35, 7019–7029. , L., McLelland, D., Lajous, M., and 9. Dugue VanRullen, R. (2015). Attention searches nonuniformly in space and in time. Proc. Natl. Acad. Sci. USA 112, 15214–15219. 10. Benedetto, A., Spinelli, D., Morrone, M.C., Spaak, E., Lange, F., de Jensen, O., Tomassini, A., Spinelli, D., Jacono, M., Sandini, G., et al. (2016). Rhythmic modulation of visual contrast discrimination triggered by action. Proc. Biol. Sci. 283, 3536–3544. 11. Kienitz, R., Schmiedt, J.T., Shapcott, K.A., Kouroupaki, K., Saunders, R.C., and Schmid, M.C. (2018). Theta rhythmic neuronal activity and reaction times arising from cortical receptive field interactions during distributed attention. Curr. Biol. 28, 2377– 2387. 12. Buschman, T.J., and Kastner, S. (2015). From behavior to neural dynamics: an integrated theory of attention. Neuron 88, 127–144. 13. Landau, A.N., Schreyer, H.M., van Pelt, S., and Fries, P. (2015). Distributed attention is implemented through theta-rhythmic gamma modulation. Curr. Biol. 25, 2332–2337. 14. VanRullen, R. (2016). Perceptual cycles. Trends Cogn. Sci. 20, 723–735. 15. Ho, H.T., Leung, J., Burr, D.C., Alais, D., and Morrone, M.C. (2017). Auditory sensitivity and decision criteria oscillate at different frequencies separately for the two ears. Curr. Biol. 27, 3643–3649.e3. 16. Tomassini, A., Ambrogioni, L., Medendorp, W.P., and Maris, E. (2017). Theta oscillations locked to intended actions rhythmically modulate perception. Elife 6, 1–18. 17. Berger, H. (1930). Uber das Elektrenkephalogramm des Menschen. Zweite Mitteilung. J. Psycho. Neurol. 40, 160–179. 18. de Almeida, L., Idiart, M., and Lisman, J.E. (2009). A second function of gamma frequency oscillations: an E%-Max winner-take-all mechanism selects which cells fire. J. Neurosci. 29, 7497–7503.
4. Landau, A.N., and Fries, P. (2012). Attention samples stimuli rhythmically. Curr. Biol. 22, 1000–1004.
19. Jensen, O., Gips, B., Bergmann, T.O., and Bonnefond, M. (2014). Temporal coding organized by coupled alpha and gamma oscillations prioritize visual processing. Trends Neurosci. 37, 357–369.
5. Fiebelkorn, I.C., Saalmann, Y.B., and Kastner, S. (2013). Rhythmic sampling within and between objects despite sustained attention at a cued location. Curr. Biol. 23, 2553–2558.
20. Spyropoulos, G., Bosman, C.A., and Fries, P. (2017). A theta rhythm in awake macaque V1 and V4 and its attentional modulation. bioRxiv, https://doi.org/10.1101/117804.
R832 Current Biology 28, R826–R845, August 6, 2018