Final Report Summary - SPATIAL ATTENTION (Neural mechanisms underlying rapid modulation of spatial attention in cortex and superior colliculus)
We are constantly inundated with a barrage of sensory information. At any given instant, however, only a small critical fraction of this sensory world is relevant for taking action. Our brains must select out the critical input and use it to guide behavioural response. We often call this selection process attention.
In 1890, William James described attention as “Everyone knows what attention is”. However, it wasn't until the 1970's, when the psychophysicist David Posner developed a task to systematically study the psychophysics of visual attention in humans, was substantial progress made on spatial attention (Posner, 1980; Posner & Cohen, 1984). He designed a covert spatial attention task; in which subjects changed the locus of their attention while other variables remained constant (e.g. eye position). This task allowed the internal orienting or “spot-light" of the brain towards a specific region in space to be isolated and studied (see Aim 1). Attention is typically divided into two categories. Attention to a locus in space evoked by an external stimulus (i.e. a rapid movement in a specific part of the visual field) is called "bottom-up" attention. Whereas attention based on an abstract non-spatial cue that signals the attention-shift to a specific locus in space is referred to as "top-down" attention. Here, we focus on "top-down" attention given that brain areas engaged in this task have been identified
The major objective of this research was developing the first spatial attention task in mice, and leverage this genetically tractable animal model to dissect the underlying microcircuit, we aim to reveal new fundamental principles controlling rapid changes sensory information processing in the brain. Additionally, objectives of this proposal are to determine the neural mechanisms in prefrontal cortex (PFC) and superior colliculus (SC) underlying the rapid routing of sensory information during spatial attention. To gain new insight and fully address this question our approach combines multiple disciplines and tiers of neuroscience.
Given that this task paradigm has substantially contributed to the field, with thousands of publications in primates, we are confident that finalizing development of this task in the mouse model, amenable to cellular and circuit tools, will generate an entirely new powerful field of research on attention.
Deliverables and Results
A. Building of three behavior rigs for head-fixed training of mice in the visual spatial attention task;
B. The behaviour task was designed and implemented using Matlab, Psychophysics toolbox and RT-Linux. Behavioral task comprised: 1) At the beginning of each trial, an auditory cue instructs the animal whether to report the change in luminance on either their upper or lower visual hemifield (high or low tone); 2) the mouse learns to initiate and trial independently, and controls visual cues on a visual monitor by running on the treadmill. This close-loop control engages animals in a manner designed to emulate the immersive environment of ‘video-game.’ Mice move the visual cue to a location on either the upper or lower hemifield which is the final locus where animal should perform a visual discrimination.
C. We developed several training protocols with both ‘bottom up’ and ‘top down’ attention tasks. Up to 15 animals are trained per day. Below is simplified version of one protocol. Nonetheless, protocols are finalized as in publications.
1. Day 1, headplate implant;
2. Day 2 - 8, water deprivation (1 mL/day);
3. Day 9, training begins, mice learn to run on a treadmill to initiate the trial and lick for water;
4. Day 10 -17, mice detect change in visual stimulus; the time window of response is adapted enabling mice to respond faster as they learn;
5. Day 18 - 21, visual and auditory negative reinforcement are introduced and mice must withhold on trials where no change in visual stimulus occur;
6. Day 22, full task, visual and auditory cues are introduced at the beginning of the trial, signaling which visual stimulus must be detected;
Notably, two of the behaviour tasks were designed and implemented using Matlab, Psychophysics toolbox and RT-Linux. The mouse learns to initiate and trial independently, and controls visual cues on a visual monitor by running on the treadmill. This close-loop control engages animals in a manner designed to emulate the immersive environment of ‘video-game.’ Mice move the visual cue to a location on either the upper or lower hemifield which is the final locus where animal should perform a visual discrimination.
D. Development of surgical window for two-photo cellular level microscopy from the SC. As part of Aim 1, we have also developed a novel surgical technique to image the SC. The SC is a midbrain structure partially occluded by the central sinus and retrosplenial cortex. We developed a technique of laser cutting custom polycarbonate pyramid shaped optical spacers. These spacers are used to delicately surgically displace the central sinus making the medial portion of the SC accessible for imaging. While optimization is ongoing, we have confirmed that this enables both macroscopic intrinsic imaging and two-photon imaging of cell bodies.
The results generated by this research demonstrate that mice can indeed learn to discriminate between distinct stimuli at different spatial locations, and can rapidly (on the time scale of minutes) alternate motor actions according to the spatial location and auditory cue presented to them. We are currently completing the implementation of the behavioral task. These will be combined with physiological and optical recordings in the prefrontal cortex (PFC) and the superior colliculus (SC), brain areas established as critical for this behavior, Two high profile publications are expected to result from this work.
Training and Transfer
This training grant contributed to the technology and scientific quality of the Champalimaud Neuroscience Programme (CNP) as well as the larger European research environment in the following specific ways: (i) brought extensive knowledge on mouse visual system and cortical circuitry by optogenetics to the CNP. Specifically, Dr. Atallah trained members of the Champalimaud Neuroscience graduate programme who are beginning to perform electrophysiological measurements in visual cortex. Dr. Atallah also continued to regularly consult and attend the lab meetings of the Petreanu and Paton Labs to provide ongoing input; (ii) brought the two-photon assisted targeted in vivo whole-cell patch-clamp recordings experience and expertise to the CNP in performing targeted in vivo whole-cell patch-clamp recordings, both ‘blind’ as well as two-photon assisted. Dr. Atallah have trained students in the Costa Lab to both built a (iii) two-photon microscope and (iv) perform two-photon imaging of neurons on this microscope.
Funding from the Marie Curie IIF was already invaluable as enabled me to produce new knowledge, chiefly by: (i) gaining the expertise in quantitative analysis of behavioural; Dr. Atallah regularly consult myself and Dr. Joseph Paton when designing the behavior tasks and perhaps more importantly the ongoing training protocol; they have taught me critical steps in this process i.e. quantitative evaluation of a broad range of factors that underlie mouse performance: movement, reward rates, i.e. contingency across trials; (ii) Dr. Atallah has been trained and developed new modules on the hardware and software platforms to support rodent behavioural training and quantification; Dr. Atallah have also contributed to two publications on this process which have been published (See Publication section).
Finally, there is no website associated to this project, and it was not foreseen at the proposal. Nonetheless Dr Atallah contacts are Fundação Champalimaud, Champalimaud Neuroscience Programme, Avenida de Brasilia, 1400-038 Lisboa, Portugal; email: Bassam.Atallah@neuro.fchampalimaud.org
In 1890, William James described attention as “Everyone knows what attention is”. However, it wasn't until the 1970's, when the psychophysicist David Posner developed a task to systematically study the psychophysics of visual attention in humans, was substantial progress made on spatial attention (Posner, 1980; Posner & Cohen, 1984). He designed a covert spatial attention task; in which subjects changed the locus of their attention while other variables remained constant (e.g. eye position). This task allowed the internal orienting or “spot-light" of the brain towards a specific region in space to be isolated and studied (see Aim 1). Attention is typically divided into two categories. Attention to a locus in space evoked by an external stimulus (i.e. a rapid movement in a specific part of the visual field) is called "bottom-up" attention. Whereas attention based on an abstract non-spatial cue that signals the attention-shift to a specific locus in space is referred to as "top-down" attention. Here, we focus on "top-down" attention given that brain areas engaged in this task have been identified
The major objective of this research was developing the first spatial attention task in mice, and leverage this genetically tractable animal model to dissect the underlying microcircuit, we aim to reveal new fundamental principles controlling rapid changes sensory information processing in the brain. Additionally, objectives of this proposal are to determine the neural mechanisms in prefrontal cortex (PFC) and superior colliculus (SC) underlying the rapid routing of sensory information during spatial attention. To gain new insight and fully address this question our approach combines multiple disciplines and tiers of neuroscience.
Given that this task paradigm has substantially contributed to the field, with thousands of publications in primates, we are confident that finalizing development of this task in the mouse model, amenable to cellular and circuit tools, will generate an entirely new powerful field of research on attention.
Deliverables and Results
A. Building of three behavior rigs for head-fixed training of mice in the visual spatial attention task;
B. The behaviour task was designed and implemented using Matlab, Psychophysics toolbox and RT-Linux. Behavioral task comprised: 1) At the beginning of each trial, an auditory cue instructs the animal whether to report the change in luminance on either their upper or lower visual hemifield (high or low tone); 2) the mouse learns to initiate and trial independently, and controls visual cues on a visual monitor by running on the treadmill. This close-loop control engages animals in a manner designed to emulate the immersive environment of ‘video-game.’ Mice move the visual cue to a location on either the upper or lower hemifield which is the final locus where animal should perform a visual discrimination.
C. We developed several training protocols with both ‘bottom up’ and ‘top down’ attention tasks. Up to 15 animals are trained per day. Below is simplified version of one protocol. Nonetheless, protocols are finalized as in publications.
1. Day 1, headplate implant;
2. Day 2 - 8, water deprivation (1 mL/day);
3. Day 9, training begins, mice learn to run on a treadmill to initiate the trial and lick for water;
4. Day 10 -17, mice detect change in visual stimulus; the time window of response is adapted enabling mice to respond faster as they learn;
5. Day 18 - 21, visual and auditory negative reinforcement are introduced and mice must withhold on trials where no change in visual stimulus occur;
6. Day 22, full task, visual and auditory cues are introduced at the beginning of the trial, signaling which visual stimulus must be detected;
Notably, two of the behaviour tasks were designed and implemented using Matlab, Psychophysics toolbox and RT-Linux. The mouse learns to initiate and trial independently, and controls visual cues on a visual monitor by running on the treadmill. This close-loop control engages animals in a manner designed to emulate the immersive environment of ‘video-game.’ Mice move the visual cue to a location on either the upper or lower hemifield which is the final locus where animal should perform a visual discrimination.
D. Development of surgical window for two-photo cellular level microscopy from the SC. As part of Aim 1, we have also developed a novel surgical technique to image the SC. The SC is a midbrain structure partially occluded by the central sinus and retrosplenial cortex. We developed a technique of laser cutting custom polycarbonate pyramid shaped optical spacers. These spacers are used to delicately surgically displace the central sinus making the medial portion of the SC accessible for imaging. While optimization is ongoing, we have confirmed that this enables both macroscopic intrinsic imaging and two-photon imaging of cell bodies.
The results generated by this research demonstrate that mice can indeed learn to discriminate between distinct stimuli at different spatial locations, and can rapidly (on the time scale of minutes) alternate motor actions according to the spatial location and auditory cue presented to them. We are currently completing the implementation of the behavioral task. These will be combined with physiological and optical recordings in the prefrontal cortex (PFC) and the superior colliculus (SC), brain areas established as critical for this behavior, Two high profile publications are expected to result from this work.
Training and Transfer
This training grant contributed to the technology and scientific quality of the Champalimaud Neuroscience Programme (CNP) as well as the larger European research environment in the following specific ways: (i) brought extensive knowledge on mouse visual system and cortical circuitry by optogenetics to the CNP. Specifically, Dr. Atallah trained members of the Champalimaud Neuroscience graduate programme who are beginning to perform electrophysiological measurements in visual cortex. Dr. Atallah also continued to regularly consult and attend the lab meetings of the Petreanu and Paton Labs to provide ongoing input; (ii) brought the two-photon assisted targeted in vivo whole-cell patch-clamp recordings experience and expertise to the CNP in performing targeted in vivo whole-cell patch-clamp recordings, both ‘blind’ as well as two-photon assisted. Dr. Atallah have trained students in the Costa Lab to both built a (iii) two-photon microscope and (iv) perform two-photon imaging of neurons on this microscope.
Funding from the Marie Curie IIF was already invaluable as enabled me to produce new knowledge, chiefly by: (i) gaining the expertise in quantitative analysis of behavioural; Dr. Atallah regularly consult myself and Dr. Joseph Paton when designing the behavior tasks and perhaps more importantly the ongoing training protocol; they have taught me critical steps in this process i.e. quantitative evaluation of a broad range of factors that underlie mouse performance: movement, reward rates, i.e. contingency across trials; (ii) Dr. Atallah has been trained and developed new modules on the hardware and software platforms to support rodent behavioural training and quantification; Dr. Atallah have also contributed to two publications on this process which have been published (See Publication section).
Finally, there is no website associated to this project, and it was not foreseen at the proposal. Nonetheless Dr Atallah contacts are Fundação Champalimaud, Champalimaud Neuroscience Programme, Avenida de Brasilia, 1400-038 Lisboa, Portugal; email: Bassam.Atallah@neuro.fchampalimaud.org