Shuttle-box avoidance learning is well-established in behavioral neuroscience. This protocol describes how shuttle-box learning in rodents can be combined with site-specific electrical intracortical microstimulation (ICMS) and simultaneous chronical in vivo recordings as a tool to study multiple aspects of learning and perception.
Shuttle-box avoidance learning is a well-established method in behavioral neuroscience and experimental setups were traditionally custom-made; the necessary equipment is now available by several commercial companies. This protocol provides a detailed description of a two-way shuttle-box avoidance learning paradigm in rodents (here Mongolian gerbils; Meriones unguiculatus) in combination with site-specific electrical intracortical microstimulation (ICMS) and simultaneous chronical electrophysiological in vivo recordings. The detailed protocol is applicable to study multiple aspects of learning behavior and perception in different rodent species.
Site-specific ICMS of auditory cortical circuits as conditioned stimuli here is used as a tool to test the perceptual relevance of specific afferent, efferent and intracortical connections. Distinct activation patterns can be evoked by using different stimulation electrode arrays for local, layer-dependent ICMS or distant ICMS sites. Utilizing behavioral signal detection analysis it can be determined which stimulation strategy is most effective for eliciting a behaviorally detectable and salient signal. Further, parallel multichannel-recordings using different electrode designs (surface electrodes, depth electrodes, etc.) allow for investigating neuronal observables over the time course of such learning processes. It will be discussed how changes of the behavioral design can increase the cognitive complexity (e.g. detection, discrimination, reversal learning).
A fundamental aim of behavioral neuroscience is to establish specific links between neuronal structural and functional properties, learning, and perception. Neural activity associated with perception and learning can be studied by electrophysiological recording of action potentials and local field potentials in various brain structures at multiple sites. Whereas electrophysiological recordings provide correlative associations between neural activity and behavior, direct electrical intracortical microstimulation (ICMS) for over a century has been the most direct method for testing causal relationships of excited populations of neurons and their behavioral and perceptual effects1–3. Many studies have demonstrated that animals are able to make use of various spatial and temporal properties of electrical stimuli in perceptual tasks depending on the stimulation site within for instance retinotopic4, tonotopic5, or somatotopic6 regions in the cortex. Propagation of electrically evoked activity in the cortex is mainly determined by the layout of axonal fibers and their distributed synaptic connectivity2 that, in cortex, is clearly layer-dependent7. The resultant polysynaptic activation evoked by ICMS is henceforth much more wide-spread than direct effects of the electrical field2,8,9. This explains why thresholds of perceptual effects elicited by intracortical microstimulation can be strongly layer-dependent8,10,11 and site-dependent9. A recent study demonstrated in detail that stimulation of upper layers yielded more wide-spread activation of corticocortical circuits in mainly supragranular layers, while stimulation of deeper layers of cortex result in a focal, recurrent corticoefferent intracolumnar activation. Parallel behavioral experiments revealed that the latter has much lower perceptual detection thresholds8. Therefore, the advantage of site-specific ICMS as conditioned stimuli was exploited in combination with electrophysiological recordings to causally relate specific cortical circuit activations8 to behavioral measures of learning and perception in the shuttle-box.
The two-way shuttle-box paradigm is a well-established laboratory apparatus to study avoidance learning12. A shuttle-box consists of 2 compartments separated by a hurdle or doorway. A conditioned stimulus (CS) that is represented by a suitable signal like a light or sound, is contingently followed by an aversive unconditioned stimulus (US), as for instance a foot shock over a metal grid floor. Subjects can learn to avoid the US by shuttling from one shuttle-box compartment to the other in response to the CS. Shuttle-box learning involves a sequence of distinguishable learning phases13,14: First, subjects learn to predict the US from the CS by classical conditioning and to escape from the US by instrumental conditioning, as the US is terminated upon shuttling. In a next phase, subjects learn to avoid the US altogether by shuttling in response to the CS before US onset (avoidance reaction). Generally, shuttle-box learning involves classical conditioning, instrumental conditioning, as well as goal-directed behavior depending on learning phase14.
The shuttle-box procedure can be set up easily and generally produces robust behavior after a few daily training sessions15–17. In addition to simple avoidance conditioning (detection), the shuttle-box can be further used to study stimulus discrimination by employing go/nogo paradigms. Here, animals are trained to avoid the US by a conditioned response (CR) (go behavior; shuttle into opposite compartment) in response to a go-stimulus (CS+) and by nogo behavior (stay in the current compartment; no CR) in response to a nogo-stimulus (CS-). Parallel microstimulation and recording of neural activity with high-density multielectrode arrays allow to study the physiological mechanisms underlying successful learning. Several technical details that are fundamental for the successful combinations of shuttle-box training, ICMS and parallel electrophysiology, will be discussed.
All experiments presented in this work were conducted in agreement with the ethical standards defined by the German law for the protection of experimental animals. Experiments were approved by the ethics committee of the state of Saxony-Anhalt.
1. Custom-made Multichannel Electrode Arrays for Microstimulation and Recording
2. Surgical Implantation of Arrays into Auditory Cortex in Anaesthetized Mongolian Herbils for Chronic Usage
3. Two-way Shuttle-box Designs Using ICMS as Conditioned Stimulus
4. In Vivo Electrophysiological Techniques in Learning Animals
5. Histological Analysis of Electrode Positions
This section illustrates a representative example of shuttle-box learning in a Mongolian gerbil. The subject was trained to discriminate the ICMS site between two stimulation electrodes implanted 700 µm apart from each other in auditory cortex (Figures 1 and 2). Stimulation arrays can be customized in different spatial designs (Figure 1). Here, discrimination of the two ICMS sites was learned within 3 training sessions with presentation of 30 CS+ and CS- each (Figure 3A-C). This is indicated by a stable significant difference of the CR rates of hit and false alarm responses throughout 7 consecutive training sessions (Figure 3B). Correspondingly, d’ is >1 in these sessions (Figure 3C). Quick escape latencies towards the US are fundamental, as they reflect an effective aversive unconditioned response. This can be guaranteed by adapting foot shock strength from 200 µA in 50 µA steps till escape latencies are short (see Figure 3E). In parallel, electrophysiological recordings from an ECoG-array allow to assess the site-specific spatiotemporal activation patterns evoked by intracortical electrical CS+ or CS- at stimulation sites separated by ~700 µm (Figure 4).
Figure 1. Electrode array designs. (A) Depth array (2 x 1) for intracortical microstimulation at two different sites in cortex. Electrodes are arranged at an interelectrode distance of ~ 700 µm. Other spatial designs can allow for layer-dependent local ICMS in different cortical depths or lateral arrays with stimulation sites along a specific axis of cortical tissue, as for instance the tonotopic gradient of auditory cortex8. (B) Epidural surface array (3 x 6) for the recording of the electrocorticogram at high spatial resolution. Electrodes were made from stainless steel wire (Ø 256 µm) arranged in a 3×6 matrix with an interelectrode distance of ~ 600 µm. Please click here to view a larger version of this figure.
Figure 2. Positioning of implanted stimulation and recording electrodes. (A) A pair of two stimulation electrodes (see Figure 1A) S1 (dark green) and S2 (light green) are implanted into the depth of the right primary auditory field AI close to its input layer IV. Electrode tips can be positioned along the rostrocaudal axis (caudal electrode S1, rostral electrode S2) with an interelectrode distance of ~ 700 µm. The 3 x 6 ECoG recording array (600 µm interelectrode distance) is centered epidurally over the right AI. (B) Nissl-stained horizontal section of respective brain region after experimental procedure shows two small lesions (arrows), which were caused by the tips of the two implanted stimulation electrodes indicating their location within temporal cortex. The position can be further evaluated by “Prussian Blue” staining. This figure has been modified from Deliano et al., 2009. Please click here to view a larger version of this figure.
Figure 3. Shuttle-box training data and analysis of one individual animal. (A) Schemata on the right describe task design for CS+ and CS- trials in a two-way shuttle-box discrimination task and behavioral outcomes. (B) Learning curves plotted as hit and false-alarm rates of individual training sessions. Significant differences between hit and false alarm rates are marked by asterisks (Games-Howell test, p<0.05). (C) Sensitivity index d’>1 (see 3.2.7) can be used as threshold criterion for successful discrimination. (D) Monitoring of the spontaneous crossings during the habituation phase generally show a decrease over sessions. (E) Response latencies during CS+ trials are plotted for individual trials over all training sessions. All responses with latencies below 6 sec correspond to successful hit responses. Note the longer escape latencies in the first half of the first session. After increasing the foot shock strength escape latencies decreased below 2 sec after US onset indicating sufficient shock control. Histograms (right inset) of response latencies are bimodal corresponding to hit responses ( < 6 sec) and escape responses (6 – 8 sec). Please click here to view a larger version of this figure.
Figure 4. Parallel electrophysiological recording in a learning animal. (A) Typical example of an electrically evoked potential (EEP) from a single animal averaged across CS+ trials in a single session of training. Data was recorded from an EcoG-array. The figure compares the EEP trace before (black) and after (red) removal of single pulse stimulus artifacts. Details of artifact reduction see section 4.1.7. The early prominent negative peak can be seen at a latency of 20 ms (N20). (B) Further analysis of the spatial distributions of the N20 amplitude in response to a CS+ at the rostral stimulation electrode (top) and to a CS- at the caudal stimulation electrode (bottom) reveal the spatial resolution of evoked states throughout auditory cortex. Anatomical directions relative to the recording array are indicated by arrows (d, dorsal; c, caudal; l, lateral; m, medial; r, rostral; v, ventral). This figure has been modified from Deliano et al., 2009. Please click here to view a larger version of this figure.
This protocol describes a method of simultaneous site-specific ICMS and multi-channel electrophysiological recordings in a learning animal by using a two-way aversive foot-shock controlled shuttle-box system. The protocol emphasizes technical key concepts for such combination and points out the importance of grounding the animal only via its common ground electrode, leaving the gridfloor at a floating voltage. Here, auditory shuttle-box learning was applied to Mongolian gerbils as learning-related plastic reorganizations of the auditory cortex in these animals have been studied extensively8,12,14,15,21,22. Nevertheless, the described protocol can be adapted with minor changes to other rodent species, as for instance mice16. In this respect it is important to consider species-specific adaptations concerning recovery time after the surgery (2.17), height of the hurdle (2.1.1), and foot-shock sensitivity of individual animals, which can be highly variable (3.1.3-3.1.6).
The protocol further gives detailed explanations on how custom-made electrode designs can be used to stimulate different sites in cortical tissue leading to distinct network activations as derived from the analysis of concurrent electrophysiological multielectrode recordings8,23. Depending on distance of electrodes one can stimulate different regions of for instance topographical maps9. By applying layer-dependent ICMS it is possible to differentially activate long-range corticocortical projections leading to more wide-spread activation of cortex by stimulation in cortical input layers III-IV. Instead, stimulation in corticoefferent output layers V-VI led to a much more focal activation of intracortical and corticothalamic feedback circuits8. When using stimulation arrays with two or more stimulation electrodes, bipolar ICMS can be applied instead of monopolar ICMS. A bipolar stimulation mode more effectively recruits neuronal fibers running parallel to the electrode tips, preferably in the direction of the cathodic pole compared with nonparallel fibers24. Such a stimulation configuration hence increases the directional specificity of the evoked neuronal network activations8. These particular direct manipulations of cortical sublaminar network activities using ICMS8,9, have so far not been shown by any other technique3. As an example of the power of this method, a recent report unraveled the contribution of cortico-thalamocortical feedback circuits to perception using detection learning of intracortical electrical stimuli8. This demonstrates that direct cortical microstimulation is an effective and state-of-the-art method to causally link activity in specified neuronal circuits and behavior1,3,11,25. By local electrical stimulation of cortical regions corresponding to specific topographic map features, as for instance a tonotopic region in the auditory cortex, subjects can be trained in transfer learning paradigms to compare properties of percepts elicited by central electrical or peripheral sensory stimulation. Such experiments might stimulate the development of stimulation strategies for sensory cortical neuroprostheses5,9. This protocol can also be employed in the electrical stimulation of other brain areas, as for instance the ventral tegmental area, to study reward processing and the neuronal underpinnings of deep brain stimulation26. Critical for effective microstimulation are several technical details that must be considered on the background of the individual setup and electrodes used. In general, influence of stimulation parameters, like stimulation amplitude, polarity, electrode orientation, etc., have been reviewed11,24. Of importance is the charge transfer by the electrode. The impedance of an electrode hence is a critical factor. Hence, check that the impedance of the electrode contacts is in the kΩ range before implantation.
Several additional phenomena of learning can be studied by appropriate variation of the described basic design. For instance, discrimination learning in contrast to simple detection learning can be investigated by introducing at least two stimuli that have to be associated with go and nogo responses, respectively14,15. Similarly category formation learning can be studied by combining such discrimination paradigms12,21. Shuttle-box paradigms also can be employed to investigate working memory, behavioral inhibition and cognitive flexibility as for instance necessary for successful reversal learning14,17 or set shifting. Working memory can be assessed by comparing 'delay' and 'trace' conditioning. In 'delay' conditioning27, the CS is presented throughout the critical CS-US time window without delay between CS offset and US onset. In 'trace' conditioning, on the other hand, there is a delay of several seconds after the offset of transient CS presentation. In contrast to 'delay' conditioning 'trace' conditioning puts a high load on working memory and cortical processing. Combining discriminative shuttle-box learning paradigms with the analysis of spatiotemporal patterns in the ongoing electrocorticogram, is a suitable method to identify dynamical states of auditory cortex related to the stimulus discrimination9, and category formation21. However, as shuttle-box training is classically used as two-way avoidance task, the general conceptual problems with avoidance learning do apply for all of these behavioral designs; namely that successful avoidance behavior explicitly prevents the occurrence of the stimulus that serves as the reinforcer. Appetitive reinforcement, for instance by direct electrical stimulation of midbrain reward circuits, has only been applied to shuttle-box learning in some studies26. Also, shuttle-box learning has mainly been used with rodent species and has been rarely applied in larger laboratory animals, as for instance dogs.
Besides with electrophysiological analysis, shuttle-box learning can be further combined with pharmacological intervention8,17, lesion techniques15, microdialysis28, or optogenetics. Especially the combination of our protocol with optogenetic tools, either by viral infection of the model system (i.e., Mongolian gerbils), or by genetically modified animals, as mice, would allow to increase in particular the cellular subtype specificity of artificial neuronal activation including cortical inhibition, which is not accessible using ICMS3.
The authors have nothing to disclose.
The work was supported by grants from the Deustche Forschungsgemeinschaft DFG and the Leibniz-Institute for Neurobiology. We thank Maria-Marina Zempeltzi and Kathrin Ohl for technical assistance.
Teflon-insulated stainless steel wire | California Fine Wire | diam. 50µm w/ isolation | |
Pin connector system | Molex Holding GmbH | 510470200 | 1.25 mm pitch PicoBlade |
TEM grid Quantifoil | Science Services | EQ225-N27 | |
Dental acrylic Paladur | Heraeus Kulzer | 64707938 | |
Hand-held drill OmniDrill35 | WPI | 503599 | |
Ketamine 500mg/10ml | Ratiopharm GmbH | 7538837 | |
Rompun 2%, 25ml | Bayer Vital GmbH | 5066.0 | |
Sodium-Chloride 0.9%, 10ml | B.Braun AG | PRID00000772 | |
Lubricant KY-Jelly | Johnson & Johnson | ||
Shuttle-box E10-E15 | Coulbourn Instruments | H10-11M-SC | |
Stimulus generator MCS STG 2000 | Multichannel Systems | ||
Plexon Headstage cable 32V-G20 | Plexon Inc. | HSC/32v-G20 | |
Plexon Headstage 32V-G20 | Plexon Inc. | HST/32v-G20 | |
PBX preamplifier 32 channels | Plexon Inc. | 32PBX box | |
Multichannel Acquisition System | Plexon Inc. | MAP 32/HLK2 | |
Cryostate CM3050 S | Leica Microsystems GmbH | ||
Signal processing Card Ni-Daq | National Instruments | ||
Lab StandardTM Stereotaxic Instruments | Stoelting Co. | ||
Audio attenator g.pah | g.pah Guger technologies | ||
Cresyl violet acetate | Roth GmbH | 7651.2 | |
Roticlear | Roth GmbH | A538.1 | |
Sodium acetate trihydrate | Roth GmbH | 6779.1 | |
Potassium hexacyanoferrat(II) trihydrate | Roth GmbH | 7974.2 | |
Di-sodium hydrogen phospahte dihydrate | Merck | 1,065,801,000 | |
ICM Impedance Conditioning Module | FHC | 55-70-0 | |
Animal Temperarture Controler | World Precision Instruments | ATC2000 |