Tactile Attention in the Vibrissal System

• Dr. Ben Mitchinson, Adaptive Behaviour Research Group, The University Of Sheffield, Sheffield, South Yorkshire, UK

Attention is a rich research area concerning itself with many aspects of brain function and animal behaviour (even as far as consciousness). This article focuses on the most prosaic aspect of attention, however, that of overt orienting of sensory receptors to the location of an 'attended stimulus' in space. Whilst overt orienting of the eyes has been studied in great detail, orienting of the vibrissal array of rodents has only recently begun to be investigated, despite the importance of this organ as an experimental model to a wide variety of investigations. This article reviews current understanding of orienting in this organ and identifies major open questions.

The expression of attention

Attention as a faculty of neural systems need not, in general, have a direct behavioural correlate (or, in principle, any behavioural correlate at all). Purported system variables associated with 'covert' attention management can be measured behaviourally only indirectly through the impact they have on subsequent stimulus-response or object search times or efficiencies (Ward 2008). Models of attentional mechanisms can be complex and multi-faceted in attempting to account for the effects of immediate sensory stimuli, prior knowledge, motivation, instruction, and so on, on the variables visible through these relatively small apertures (Frintrop et al. 2010). These models have been developed using data from visual system studies (in models such as cat and primate); attention is often studied using human participants since the required manipulations are often not invasive (Posner 1980).

Despite the complexity of the evolution of these internal states, however, the 'overt' orienting of the sensory organs, the head, and the body, to focus sensory apparatus onto a selected region of space (the 'orienting reflex') is both a rather more straightforward expression of attention switching and absolutely central to everyday behaviour (Itti and Koch 2001). In primates, orienting is typically led by the visual system's fovea and is modelled as a series of discrete 'saccades' of the visual fixation point from one spatial location to the next (Itti et al. 1998). This behaviour is predictable, and has been characterised in a range of tasks including, for instance, scene search (Yarbus 1967), reading (Rayner 1998) and locomotion (Hollands et al. 1995). The contemporary ease of measuring eye movements in a human subject facilitates the use of this model (Duchowski 2007).

Rodent models of attention

Whilst observation of primate visual systems provides relatively controlled access to attentional mechanisms, and particularly to spatial variables, the use of rodent models may be desirable for some studies, for example if the goal is drug discovery or if the manipulation is otherwise invasive. Techniques for observing attentional mechanisms in rodents are less comprehensive, but a variety of choice/reward protocols have been employed which make use of the spatial component of the apparatus only for the indication of the choice (Sagvolden et al. 2005). One well-established paradigm is the 5-choice serial reaction time task (5CSRTT) (Robbins 2002) which comprises the measurement of the procedure and timing of nose-poking into apertures signalled by (paradigmatically, visual) stimuli. Results from the 5CSRTT are described as indicating characteristics such as 'impulsivity' and 'compulsivity' (Robbins 2002, Sanchez-Roige et al. 2012). The potential relevance of this paradigm to research into contemporary medical problems such as attention-deficit disorders is clear (in fact, the task was originally designed with just such investigations in mind).

The sensory modality upon which rodents rely the most is, under many conditions, the tactile whisker (vibrissa) system rather than vision (see Vibrissal behavior and function). One consequence of that in this context follows from the observation that 'orienting' of the tactile whisker sensory organ to a spatial location implies the bringing of the organ to, rather than just the pointing of the organ towards, the target of attention, a result of the fundamentally limited range of tactile sensing (Mitchinson and Prescott 2013). Thus, the animal's morphology dictates a convergence between sensory orienting and 'taking action' (which might include, for example, the orienting of motor resources such as the mouth and locomotion to a location from where action can be taken). Investigations using Whiskered robots have demonstrated that linking a changing focus of spatial attention to orienting of the snout generates plausible sequences of gross behaviour (Mitchinson et al. 2012, Mitchinson and Prescott 2013, Prescott et al. in press). This convergence has previously been described using the expression orienting is acting (Mitchinson in press). Note that this convergence is present in all systems to an extent even though it is much less marked when it is remote sensors that are being oriented.

Meanwhile, whilst many animals can move their eyes within their heads, so many whiskered animals—certainly, rodents—appear able to reposition their whiskers about their heads (Sofroniew and Svoboda 2015). Most models to date of the control of these movements have been based fairly directly on measured relationships between whisker movements and immediate sensory and motor variables rather than on central mechanisms (Towal and Hartmann 2006, Mitchinson et al. 2007, Grant et al. 2009, Deutsch et al. 2012, Arkley et al. 2014), but an analogy between these movements and saccadic movements of the eyes (which are defined in physical space) has been drawn more than once (Towal and Hartmann 2006, Pearson et al. 2007, Mitchinson et al. 2014, Sofroniew and Svoboda 2015). Taken together, these observations suggest an identification between the control of the whiskers and snout and the control of the eyes and head, both driven by the animal's changing focus of spatial attention.

A model of spatial attention mediated by vibrissae

This identification is supported, also, by the nature of observed whisker movements, and one model of whisker movement control has been proposed that makes this identification explicit, having at its core a state representing the region of space attended by the animal (Mitchinson and Prescott 2013). The model provides a succinct explanation for several observations of whisker movement and makes predictions as to how direct manipulation of the animal's spatial attention will be reflected through those movements. Experimental implementation of the model expresses rodent-like exploratory behaviour as a 'series of orients' analogous to the 'series of saccades' model of visual orienting behaviour, underlining both this analogous relationship and also the connection in the case of the rodent model with gross behaviour.

Description of the model

Figure 1: Summary of model of expression of rodent spatial attention. The locus of spatial attention is driven by sensory signals and, in turn, drives movement of the whiskers and the head. Specific observations that have been reported (CIA, HTA, see text) can be understood as correlations between states of this model. Taken from (Mitchinson and Prescott 2013).

Figure 2: Still from video of simulated rat behaviour comprising plane model of rat snout, one row of whiskers, and stationary flat obstacle. Contact-induced asymmetry is seen, as the whiskers on the right are protracted more than those on the left. Taken from (Mitchinson and Prescott 2013).

In brief, the model dictates that whiskers will be moved so as to bring them to bear on attended objects. Thus, its central variable is a representation of the immediate region of space attended by an animal, which we can refer to as the locus of spatial attention or the 'attended region' (Figure 1). This region may be compact or extensive, though 'objects' are not explicitly delineated (a distinction is usually drawn between location-based and object-based attention).

Behavioural studies have indicated that the movement of all of the whiskers on each side of the face can be well-summarised by only two variables (Grant et al. 2009); however, such a summary does not account for the shaping of the 2D surface formed by the whisker tips (Huet and Hartmann 2014). Accordingly, a fixed transform, which may be arbitrarily non-linear, maps from the representation of the attended region to the maximum protraction angle of each of the whiskers individually. This transform is chosen so that each whisker tends to be driven forward into the attended region by a small amount. A separate model component generates the rhythmic 'whisking' motion characteristic of this system (see Whisking kinematics and Whisking pattern generation) so that the instantaneous protraction angle of each whisker is derived by combining the outputs of these two systems, and whiskers are palpated in and out of the attended region.

Meanwhile, the movement of the snout of the animal (specifically, the positioning of a generalised sensory ‘fovea’ around the mouth) is driven by the same representation on a longer timescale. In this model, whisker and head movements are both, therefore, generated as overt expressions of this attended region. Interactions between the animal and its environment then modulate the locus of spatial attention through the sensory systems; in particular, contacts between the whiskers and objects in the environment excite attention directly.

Key results

The model has been used to simulate a range of behavioural experimental paradigms and has generated results in each case that are qualitatively and quantitatively similar to those reported in animals. First, 'Head-Turning Asymmetry' (HTA) is a relationship that has been observed in rat, mouse, and opossum (Towal and Hartmann 2006, Mitchinson et al. 2011). Turning of the head is preceded by asymmetry in the protraction angles of the whiskers on each side of the head such that the whiskers appear to 'lead' the head movement. Second, 'Contact-Induced Asymmetry' (CIA, Figure 2) is a relationship between whisker-environment contact and protraction angle asymmetry (Mitchinson et al. 2007, Mitchinson et al. 2011, Deutsch et al. 2012). Contact with the environment on one side only leads to less/more protraction on the ipsilateral/contralateral side. Third, 'spread reduction' (SR) denotes the finding that the spread between the protraction angles of whiskers on each side is reduced during environmental contact (Grant et al. 2009, Huet and Hartmann 2014).

Discussion

Each of the observations listed can be alternatively interpreted as indicating a single direct reactive mechanism. However, the model based on spatial attention is sufficient to produce all of them from a single, albeit more complex, mechanism. Moreover, the attention-based model can explain anticipatory modulations of whisker movement, which reactive mechanisms cannot, and there are now many observations of these (Sachdev et al. 2003, Berg and Kleinfeld 2003, Mitchinson et al. 2007, Grant et al. 2009, Arkley et al. 2014, Voigts et al. 2015).

Direct manipulation of the animal's attention will be required to make a direct test of the model, and this experiment has not yet been performed. Given the potential for confound with direct influences on whisker movement, manipulating attention using a non-tactile modality may be the preferred experimental approach (audio or olfactory stimuli might be suitable, for instance).

Methodological potential

Methodologically, the measurement of whisker motions has become far easier over recent years primarily owing to the increased availability of high-speed and high-sensitivity videography (Diamond and Arabzadeh 2013) (coincidentally, eye movement tracking has become more accessible over much the same period, though for different reasons). As a result, measurement of whisker motion has become a potentially-useful general tool for accessing the animal's state, apart from its use in the analysis of whisker operation per se (for instance, see (Wolfe et al. 2011)). Whilst whisker motions can now be measured even in freely-behaving preparations (Arkley et al. 2014), they are available in the head-fixed preparation also (which would stand in the way of choice protocols such as 5CSRTT), and measurement can be largely automated in both cases (see Clack et al. 2012 and references therein).

Potentially, then, whisker motion measurement provides a window onto attentional mechanisms in conditions that are highly flexible and amenable to direct neural measurements and interventions. The establishment of direct connections between whisker movements and hidden attentional state variables would offer a powerful new tool for the study of attention in a rat model. Such a tool would provide access to states not available through established paradigms such as the 5CSRTT and, since whisker movements are known to respond to sensory events on very short timescales (10-15 ms, Mitchinson et al. 2007), with good temporal as well as spatial resolution. The degree to which vibrissal attention would share mechanisms and resources in rodents with other modalities would be an open question, but cross-modal effects on attention are very well established, at least, in humans (Driver and Spence 1998).

In context

If a connection between attention and whisker movement can be validated, it will have two impacts. First, as discussed, by providing a minimally-invasive tool for accessing attentional states in the behaving animal. Second, by providing a more complete understanding of how the whiskers are used by the animal to serve the purpose of Active tactile perception. Through the relationship with gross behaviour, it is perhaps particularly clear in animals that rely so heavily on their whiskers how closely these two concepts—attention and active sensing—are linked.

Given the importance of this system to many areas of neuroscience, a complete understanding of how the whiskers are moved in response to exogenous and endogenous influences would be very valuable. Indeed, experimental paradigms whereby whisker stimulation occurs as a result of whisker movement, rather than of stimulator movement, are necessary if the decoding of sensory signals is to be understood (see Vibrissal location coding).

Any commonalities identified between these two popular models (visual and vibrissal) would, per se, be a great step forward in terms of synthesis. Lines of enquiry in this regard could include investigations of whisker sensory and motor characteristics of brain areas involved in the management of spatial attention and the generation of spatially-targeted behaviours including visual orienting. An obvious example is superior colliculus, which is a well-studied area in general (May 2006, Gandhi and Katnani 2011) and has generated a slew of promising results in studies of the vibrissae both historically and more recently (see Bezdudnaya and Castro-Alamancos 2014 for a recent contribution, also see references in Mitchinson and Prescott 2013).

Intriguingly, though, it may be the differences between these systems that has the potential to be most revealing (Mitchinson and Prescott 2013). There is both neurophysiological (Benedetti 1991) and morphological (Huet and Hartmann 2014) evidence that whisker sensory signals are brought into register with visual signals in rat colliculus, a result which is itself unsurprising (Stein and Stanford 2008). Bringing into register sensory signals from modalities with mobile sensory organs requires dynamic remapping, however. In the case of the whiskers, which constantly move back and forth through space during active sensing, this remapping must occur at multiple cycles each second, a fact which may help to highlight the underlying mechanism. Meanwhile, orienting of the eyes—a remote sensor—requires the control only of azimuth and elevation, and representations of space in colliculus are exclusively described as being two-dimensional (Itti and Koch 2001) (very likely, it seems, correspondingly). In contrast, the control of just the snout (let alone the whiskers) requires a richer representation of space; on the face of it, all three dimensions must be represented, since the snout must be directed to not just towards the target of attention (Mitchinson and Prescott 2013). Understanding how these two systems, primate and rodent, correspond in these regards is bound, it seems, to be instructive.

References

• Arkley, K; Grant, R A; Mitchinson, B and Prescott, T J (2014). Strategy change in vibrissal active sensing during rat locomotion. Current Biology 24(13): 1507-1512.

• Benedetti, F (1991). The postnatal emergence of a functional somatosensory representation in the superior colliculus of the mouse. Developmental Brain Research 60(1): 51-57.

• Berg, R W and Kleinfeld, D (2003). Rhythmic whisking by rat: retraction as well as protraction of the vibrissae is under active muscular control. Journal of Neurophysiology 89(1): 104-117.

• Bezdudnaya, T and Castro-Alamancos, M A (2014). Neuromodulation of whisking related neural activity in superior colliculus. The Journal of Neuroscience 34(22): 7683-7695.

• Clack, N G et al. (2012). Automated tracking of whiskers in videos of head fixed rodents. PLoS Computational Biology 8(7): e1002591.

• Deutsch, D; Pietr, M; Knutsen, P M; Ahissar, E and Schneidman, E (2012). Fast feedback in active sensing: touch-induced changes to whisker-object interaction. PloS ONE 7(9): e44272.

• Diamond, M E and Arabzadeh, E (2013). Whisker sensory system-from receptor to decision. Progress in Neurobiology 103: 28-40.

• Driver, J and Spence, C (1998). Cross-modal links in spatial attention. Philosophical Transactions of the Royal Society of London B: Biological Sciences 353(1373): 1319-1331.

• Duchowski, A (2007). Eye Tracking Methodology: Theory and Practice, Vol. 373. Springer Science & Business Media.

• Frintrop, S; Rome, E and Christensen, H I (2010). Computational visual attention systems and their cognitive foundations: A survey. ACM Transactions on Applied Perception (TAP) 7(1): 6.

• Gandhi, N J and Katnani, H A (2011). Motor functions of the superior colliculus. Annual Review of Neuroscience 34: 205.

• Grant, R A; Mitchinson, B; Fox, C W and Prescott, T J (2009). Active touch sensing in the rat: anticipatory and regulatory control of whisker movements during surface exploration. Journal of Neurophysiology 101(2): 862-874.

• Hollands, M A; Marple-Horvat, D E; Henkes, S and Rowan, A K (1995). Human eye movements during visually guided stepping. Journal of Motor Behavior 27(2): 155-163.

• Huet, L A and Hartmann, M J (2014). The search space of the rat during whisking behavior. The Journal of Experimental Biology 217(18): 3365-3376.

• Itti, L and Koch, C (2001). Computational modelling of visual attention. Nature Reviews Neuroscience 2(3): 194-203.

• Itti, L; Koch, C and Niebur, E. (1998). A model of saliency-based visual attention for rapid scene analysis. IEEE Transactions on Pattern Analysis and Machine Intelligence 20(11): 1254-1259.

• May, P J (2006). The mammalian superior colliculus: laminar structure and connections. Progress in Brain Research 151: 321-378.

• Mitchinson, B (in press). Attention and orienting. In: T J Prescott and P F M J Verschure (Eds.), Living Machines: A Handbook of Research in Biomimetic and Biohybrid Systems. Oxford University Press.

• Mitchinson, B et al. (2011). Active vibrissal sensing in rodents and marsupials. Philosophical Transactions of the Royal Society B: Biological Sciences 366(1581): 3037-3048.

• Mitchinson, B; Martin, C J; Grant, R A and Prescott, T J (2007). Feedback control in active sensing: rat exploratory whisking is modulated by environmental contact. Proceedings of the Royal Society B: Biological Sciences 274(1613): 1035-1041.

• Mitchinson, B; Pearson, M J; Pipe, A G and Prescott, T J (2014). Biomimetic tactile target acquisition, tracking and capture. Robotics and Autonomous Systems 62(3): 366-375.

• Mitchinson, B; Pearson, M; Pipe, A and Prescott, T. (2012). The emergence of action sequences from spatial attention: Insight from rodent-like robots. In: Biomimetic and Biohybrid Systems, Vol. 7375 of Lecture Notes in Computer Science (pp. 168-179). Springer.

• Mitchinson, B and Prescott, T J (2013). Whisker movements reveal spatial attention: a unified computational model of active sensing control in the rat. PLoS Computational Biology 9(9): e1003236.

• Pearson, M J; Pipe, A G; Melhuish, C; Mitchinson, B and Prescott, T J (2007). Whiskerbot: A robotic active touch system modeled on the rat whisker sensory system. Adaptive Behavior 15(3): 223-240.

• Posner, M I (1980). Orienting of attention. Quarterly Journal of Experimental Psychology 32(1): 3-25.

• Prescott, T J et al. (in press). The robot vibrissal system: Understanding mammalian sensorimotor co-ordination through biomimetics. In: P Krieger and A A Groh (Eds.), Sensorimotor Integration in the Whisker System. Springer.

• Rayner, K (1998). Eye movements in reading and information processing: 20 years of research. Psychological Bulletin 124(3): 372.

• Robbins, T (2002). The 5-choice serial reaction time task: Behavioural pharmacology and functional neurochemistry. Psychopharmacology 163(3-4): 362-380.

• Sachdev, R N; Berg, R W; Champney, G; Kleinfeld, D and Ebner, F F (2003). Unilateral vibrissa contact: changes in amplitude but not timing of rhythmic whisking. Somatosensory & Motor Research 20(2): 163-169.

• Sagvolden, T; Russell, V A; Aase, H; Johansen, E B and Farshbaf, M (2005). Rodent models of attention-deficit/hyperactivity disorder. Biological Psychiatry 57(11): 1239-1247.

• Sanchez-Roige, S; Pena-Oliver, Y and Stephens, D N (2012). Measuring impulsivity in mice: the five-choice serial reaction time task. Psychopharmacology 219(2): 253-270.

• Sofroniew, N J and Svoboda, K (2015). Whisking. Current Biology 25(4): R137-R140.

• Stein, B E and Stanford, T R (2008). Multisensory integration: current issues from the perspective of the single neuron. Nature Reviews Neuroscience 9(4): 255-266.

• Towal, R B and Hartmann, M J (2006). Right-left asymmetries in the whisking behavior of rats anticipate head movements. The Journal of Neuroscience 26(34): 8838-8846.

• Voigts, J; Herman, D H and Celikel, T (2015). Tactile object localization by anticipatory whisker motion. Journal of Neurophysiology 113(2): 620-632.

• Ward, L M (2008). Attention. Scholarpedia 3(10): 1538. http://www.scholarpedia.org/article/Attention.

• Wolfe, J; Mende, C and Brecht, M (2011). Social facial touch in rats. Behavioral Neuroscience 125(6): 900.

• Yarbus, A L (1967). Eye Movements and Vision. New York: Plenum Press.

Sponsored by: Prof. Tony J. Prescott, Dept Psychology, Univ of Sheffield, UK

Reviewed by: Prof. Ehud Ahissar, Department of Neurobiology, The Weizmann Institute, Rehovot

Reviewed by: Dr. Goren Gordon, Personal Robots Group, Media Lab, MIT, Cambrdige, MA, United States

Accepted on: 2015-03-13 22:22:25 GMT