In press in “The New Visual Neurosciences”, MIT Press.

Expected publication date: 2013

Title

Neural Mechanisms of Target Selection in the Superior Colliculus

Authors

Uday K. Jagadisan1,4

Neeraj J. Gandhi1,2,3,4

Affiliations

Departments of Bioengineering1, Otolaryngology

2, Neuroscience

3, and Center for the Neural

Basis of Cognition (CNBC)4

University of Pittsburgh, PA 15213

Abbreviated title

Target Selection in the Superior Colliculus

Corresponding author

Neeraj J. Gandhi, Ph.D.

University of Pittsburgh

Ear and Eye Institute, Room 108

203 Lothrop St

Pittsburgh, PA 15213

Phone: 412-647-3076

FAX: 412-647-0108

E-mail: neg8@pitt.edu

Total number of words in the manuscript

10,007 (7,675 without references)

Total number of figures

3

Keywords: saccade, fixation, smooth pursuit, motor preparation, priority, Bayesian

1

Introduction

The visual environment is filled with objects that contain potentially useful information for the

survival of an organism. In order to best extract this information, the animal needs to orient itself

to the part of the visual world most relevant for its immediate behavior while ignoring the

unwanted parts. This is especially true for foveating animals such as humans and other primates,

where only a small part of the retina (the fovea) is able to resolve the visual world with greatest

detail. The brain is therefore faced with a sampling problem - how to decide, amongst the visual

clutter, where to look next? In other words, how does the selection of one or more targets for

future foveation emerge in the visual-oculomotor system?

A conceptual approach to this problem is illustrated by the following example. Imagine that you

are searching for your keys that were earlier misplaced in your room. You know what your keys

look like, and the search ends when you find an object that matches the mental image. You begin

the search by casting your glance at a random location, bringing a new set of objects into your

visual field. The new visual stimuli are not homogeneous in either their intrinsic properties

(brightness, color) or their relevance (likely location for keys). During the process of evaluating

the stimulus at the current location and deciding where to look next, your attention can be

captured by salient stimuli such as a flashing or ringing phone, a bird flying past your window,

or bright and shiny objects such as your computer screen or a bunch of loose change. You also

implicitly give weight to locations where you are more likely to have left the keys, such as your

desk or your jacket as opposed to under the bed. Moreover, you are less likely to look again in a

place you just spent some time focusing on - an exercise in futility. All these attributes of the

visual objects in your room and their respective locations must be tied together on a continual

basis to enable a thorough evaluation of where to look next. Because gaze is localized to a single

location at any given time, the competition between multiple locations must be resolved before a

decision selecting the next target is made. Finally, this evolving decision must be relayed to

premotor structures to actuate the gaze shift. Although the description of this example goes

through a series of stages, it is also possible that they are implemented in a parallel fashion in the

ongoing activity in the brain.

Numerous studies in the past couple of decades have attempted to uncover the neural correlates

of such visual target selection. Although early studies focused on the role of the fronto-parietal

cortical network, including the frontal eye fields (FEF)(Schall, 2001) and lateral intra-parietal

area (LIP)(Bisley & Goldberg, 2010), more recent work has elucidated the role of the superior

colliculus (SC) in selecting targets for eye movements (Krauzlis, Liston, & Carello, 2004). We

have parsed this wealth of information into this chapter as follows. The first section presents a

brief overview of SC, its anatomical and physiological properties, and its role in the control of

gaze. In the next section, we review target selection in the SC specifically in the context of visual

search. In the third section, we discuss the role of SC in selecting targets in other domains,

including smooth pursuit, non-visual inputs, and non-oculomotor effectors, pointing to a more

2

general function for the midbrain structure. In the fourth section we present an interpretation of

any gaze state as target selection and look at the role of SC in fixation and de-fixation. This is

followed by a discussion of motor preparation and its relation to target selection-related activity

in the SC. Finally, we conclude by discussing a framework that interprets activity in the SC

network as a priority map for action.

Anatomical and physiological properties of the SC

The SC (and its homologue in non-mammals - the optic tectum, or OT) is an evolutionarily

ancient structure whose main function seems to be to direct or orient the attention of an animal,

primarily by controlling its gaze. Located at the roof (tectum) of the brainstem, the SC can be

anatomically divided into seven distinct layers (for in depth reviews, see Huerta & Harting,

1984; Isa & Hall, 2009; May, 2006; Sparks & Hartwich-Young, 1989). These layers can be

further classified based on their physiological and functional properties into two levels. The

superficial layers (SCs) receive inputs directly from the retina as well as primary and extrastriate

visual cortices (V1, V2, V4). The pretectum and parabigeminal nucleus (the cholinergic isthmic

nucleus in non-mammals) also project to SCs. SCs in turn projects ventrally to the intermediate

and deep layers in the SC, the lateral geniculate nucleus (LGN) in the thalamus, and reciprocally

to the pretectal and parabigeminal nuclei. The intermediate and deep layers (SCid; henceforth

just called intermediate layers) receive inputs from the superficial layers, the (dorsal) fronto-

parietal cortical network involved in the processing of visuo-spatial information, including the

lateral intraparietal area (LIP) and the frontal eye fields (FEF), the dorso-lateral prefrontal cortex

(dlPFC), and the infero-temporal cortex (IT) in the ventral visual pathway. Furthermore, the

basal ganglia also project to the SCid, primarily in the form of GABA-ergic projections from the

substantia nigra pars reticulata (SNpr). There is also evidence that the SCid is the recipient of

cholinergic inputs from the parabrachial nucleus in the pons. It also receives information related

to non-visual modalities. In turn, the intermediate layer neurons project to brainstem nuclei

involved in the control of gaze - the mesencephalic reticular formation (MRF, vertical

component of gaze), and the paramedian pontine reticular formation (PPRF, horizontal

component of gaze) (Moschovakis, Scudder, & Highstein, 1996). Finally, the intermediate layer

neurons are also part of an interlaminar network that ascend back onto the superficial layers as

well as providing feedback projections to the FEF via the mediodorsal thalamus (Sommer &

Wurtz, 2004) and to the parietal and temporal cortices via the pulvinar (Berman & Wurtz, 2010;

Clower, West, Lynch, & Strick, 2001).

Another important property of the SC network, across layers, is that each colliculus represents a

topographic map of the animal’s contralateral hemifield in a retinotopic reference frame (see

review by Gandhi & Katnani, 2011b). Accordingly, neurons in the superficial layers exhibit

responses that are time-locked to the onset of visual stimuli (visual neurons) in their receptive

field. Neurons in the intermediate layers fire in response to gaze shifts (motor neurons) into their

3

movement field, or both to a gaze shift and visual stimulus onset in their response field

(visuomotor neurons). Each class of neurons exhibits a finer spectrum of activity depending on

whether their responses are phasic (transient), tonic (sustained), or a combination of the two (for

a brief overview of the various response types, see Figure 1 in McPeek & Keller, 2002). The

response field locations, to a large extent, overlap across the two layers as one proceeds dorso-

ventrally through the SC. There is also a spectrum of response types within each class of

neurons, including purely phasic, phasic-tonic, tonic and build-up like activity. The topography

on the SC tissue is as follows. The rostral half of the SC maps onto the central visual field or

small stimulus eccentricities and gaze shift amplitudes. The caudal half maps onto the peripheral

visual field - eccentric locations and large movements. Similarly, the lateral halves of the

colliculi represent the downward hemifield while the medial halves represent the upward

hemifield. The mapping from visual space to SC tissue space is non-linear, and in fact

logarithmic, along the rostro-caudal axis such that significantly more neural tissue is dedicated to

the central field compared to the extremities.

This rich diversity in the physiological properties of the SC, along with its anatomical location,

supports the idea that it is a critical node in the process of integrating sensory information to

produce overt orienting behavior (e.g., saccades). In the following sections, we will look at the

neural signatures that identify stimulus selection.

Target selection in visual search

Since as early as the first studies on the SC, it has been known that stimulus-induced activity on

the SC map is modulated by whether the stimulus is the target of an upcoming saccade or not

(Goldberg & Wurtz, 1972; Mohler & Wurtz, 1976). The “visual enhancement” is seen even in

the superficial layers (SCs) that don’t explicitly show movement related activity. Although the

effect was originally attributed to peripheral attention, it provided the first indication of a

potential role for the SC in target selection. Nevertheless, these results are limited in their

interpretability in terms of target selection because the task involved just a single visual stimulus.

In order to truly study target selection, it is important that multiple stimuli be present on the

screen at the same time.

How is the next target chosen from a field of multiple alternatives? Studies focusing on the SC

have used an “oddball visual search” paradigm to probe this question (Kim & Basso, 2008;

McPeek & Keller, 2002; Shen & Paré, 2007). The task is fairly simple - a unique visual stimulus

embedded in a search array of any number of identical distractors serves as the saccadic target.

Various groups have used a similar paradigm to study the relationship between stimulus

selection and saccadic target or response selection in the FEF and LIP as well (Schall & Hanes,

1993; Thomas & Paré, 2007; Thompson, Hanes, Bichot, & Schall, 1996). In the SC, this task

reveals a spectrum of physiological responses across the population, as reported by McPeek and

4

Keller (2002). Neurons in SCs exhibit no differential activity depending on whether a target

stimulus is in their receptive field or a distractor. In contrast, neurons in SCid discriminate the

target from distractors to varying degrees depending on their functional properties. Visuo-

movement neurons with a biphasic visual response (such as in Figure X.1B, left) display

enhanced activity in the second peak when a stimulus is the target, and this differential encoding

of the target relative to distractors persists during the prelude activity that follows the transient

burst. It has therefore been hypothesized that the first peak of the biphasic response reflects

direct sensory input whereas the second peak reflects recurrent processing of the initial activity

and/or input from top-down sources. For this subset of SCid neurons, the time at which neural

activity “selects” the target is independent of movement selection time, highlighting their role in

the conceptual stage of target selection. In other visuo-movement neurons and purely motor

neurons, in contrast, differential encoding of target and distractors occurs after the visual burst

and at a time that is correlated with saccade latency, pointing to their role in movement

preparation (McPeek & Keller, 2002).

Further insight into the precise nature of the target selection signal comes from studies recording

simultaneously from multiple sites on the SC map. Kim and Basso (2008) recorded the activity

of neurons corresponding to the four stimulus locations in an oddball visual search task. Because

the visual stimuli were constrained by electrode placement in the SC and not vice versa, the

asymmetric nature of the search array produced sessions with varying levels of performance

accuracy. The key interpretation is that instead of activity at the target site alone, the level of

discriminability between target and distractor-related activity across the entire SC predicted the

monkey’s performance on individual trials. Furthermore, the discriminability evolved with time

from onset of the search array, echoing the accumulation of a decision-like signal reported in

numerous studies and psychophysical models (Hanes & Schall, 1996; Kim & Basso, 2008;

Ratcliff, Cherian, & Segraves, 2003; Reddi, Asrress, & Carpenter, 2003). Consistent with the

idea of target selection based on a relative or competitive signal, lidocaine- or muscimol-induced

inactivation of SC selectively affects performance when the target is part of a search array

compared to when presented as an isolated stimulus (McPeek & Keller, 2004). The saccades

made under inactivation on single target trials, although slower, were always to the correct (and

only) target location. In contrast, when the target was the singleton in an oddball task,

inactivation of the SC region corresponding to the target led to a significant proportion of errors

to the irrelevant distractors. The effect was target and location-specific in that performance was

unaffected when a distractor was in the inactivation field. These results highlight the competitive

nature of target selection when multiple stimuli are vying for future foveation.

Is target selection purely a posteriori, i.e., do selection mechanisms kick in only after all the

necessary information has been presented to the visual system? Basso and Wurtz (1998) showed

that activity in the SC network is modulated by contextual information about target probability,

by systematically varying the number of possible locations at which a single target could appear

5

on a given trial. Neurons with low level pre-motor activity had higher activation when the

number of potential locations was low, and this was correlated with a reduction in saccade

reaction times. As seen before, neurons with these characteristics are also modulated by

information about target/distractor identity in visual search (McPeek & Keller, 2002). However,

activity in the phasic visual-motor burst neurons was unmodulated by target probability. Thus, a

“preparatory set” or “motor set” in the SC population activity indicates that at least part of the

selection signal can arise from prior information and can enable a faster implementation of the

selection decision (Basso & Wurtz, 1998).

Thus far, we have looked at cases where the location of the target also serves as the eventual

saccade goal. This introduces a potential confound between purely target selection related

activity and activity in preparation of the upcoming movement, especially in the SCid which is

thought to encode both processes. Observing a dissociation of target selection time from saccadic

reaction time is by itself insufficient in resolving the confound. A better way to tease apart the

two processes is to employ a task design where the saccade endpoint is spatially removed from

the stimulus cueing the movement - the antisaccade task has been widely used to do precisely

this (Juan, Shorter-Jacobi, & Schall, 2004a; Munoz & Everling, 2004; Sato & Schall, 2003). In

this task, the subject is required to saccade to an object or location diametrically opposite to a cue

stimulus that is presented alone or embedded in a search array. Everling and colleagues looked at

activity in the SC following a central instruction cue that indicated to the animal whether the trial

required a pro- (towards the peripheral stimulus) or an anti-saccade (Everling, Dorris, Klein, &

Munoz, 1999). Stimulus-related activity, in response to the same visual stimulus, was lower on

anti- trials than on pro- trials. Moreover, the activity of neurons in the rostral part of SC – those

that respond to the fixated instructional cue – was also modulated; they were higher on anti-

trials. A putative explanation for these observations is that the SC receives stronger inhibition

from top-down areas, e.g., directly from the prefrontal cortex or via the basal ganglia, to prevent

unwanted triggering of stimulus-induced prosaccades on anti-saccade trials (Everling & Munoz,

2000; Munoz & Everling, 2004). We will revisit the idea of motor preparation and the

antisaccade task in a following section.

Target selection in other domains

Do the aforementioned properties of SC extend to scenarios beyond a static visual scene, such as

when the target is in motion (e.g., smooth pursuit, looming stimulus) or where the selection is

effected by means other than a gaze shift (e.g., reach, locomotion)? In this section we review

evidence that points to a more general function of the SC in target selection.

During natural behavior, our gaze shifts are not limited to saccades or coordinated eye-head

movements. The presence of a moving stimulus can also lead to smooth pursuit behavior. Hence,

closely linked to the notion of selecting a static target for orientation is the notion of selecting a

6

moving target for tracking. It is possible that the selection of targets for smooth pursuit is

coordinated with the network that participates in target selection for saccades. An early set of

studies (Carello & Krauzlis, 2004; Krauzlis & Dill, 2002) used a behavioral task in which two

potential targets appeared at diametrically opposite locations in the visual field. These stimuli

could either remain stationary or begin to move horizontally towards the midline of the display,

and the animal was required to orient to the stimulus cued by the fixation point. Neural activity

in the SCid exhibited selectivity for the target identified by the cue, but presumably only for the

duration the target remains in the retinotopically defined receptive field of the neuron.

Furthermore, subthreshold microstimulation of SCid during the same task biased the selection of

the target contralateral to the stimulus site. In both experiments, target selection was independent

of whether the required movement was a saccade or pursuit. Nummela and Krauzlis (2011) used

a modified paradigm, in which the two targets move along directions that allow for averaging

behavior during initial pursuit. Under normal conditions, the initial pursuit traces an

approximately equally weighted average of the two target velocities. Inactivation of the SC

strongly biased this weighting against the target corresponding to the inactivated site. The results

mirror those observed for the selection of stationary targets, providing evidence for a causal role

for SC in selecting “targets” on a more general level.

In a series of studies in the barn owl, Knudsen and colleagues demonstrated the role of the optic

tectum (OT) in the selection of stimuli based on relative salience (Mysore, Asadollahi, &

Knudsen, 2011; Mysore & Knudsen, 2011a, 2011b). The OT is the non-mammalian homologue

of the SC and contains topographic representations of auditory as well as visual space (Knudsen,

1982). In one experiment, barn owls were presented with visual stimuli looming in at various

speeds (naturalistic stimuli for birds) within the receptive field of an OT neuron. The loom

stimuli were presented either alone or in the presence of a competitive distractor at another

location. One class of OT neurons was shown to represent relative salience – there is a gradual

reduction in their responses to the loom stimulus as the strength (loom speed) of the distractor is

varied. Other neurons exhibit switch-like responses when the stimulus in their receptive field

becomes the stronger of the two stimuli. Similar experiments with combined auditory and visual

stimuli showed that the switch neurons represent the strongest stimulus regardless of the

modality of the distractor stimulus (strength of auditory stimulus was modulated by the intensity

of broadband noise). Further, the switch-like property is also seen in the population activity, and

this code changes as a function of the strength of the strongest stimulus, indicating that the OT

may be involved in computing a categorical yet flexible representation of salience (Mysore &

Knudsen, 2011a).

Effects of SC inactivation on target selection for reaching movements have been reported to be

similar to the effects on target selection for saccades. Muscimol-induced inactivation of a

location on the SC map produces deficits in selecting the corresponding stimulus as a target for

reach, even in the absence of perceptual or motoric deficits (Song, Rafal, & McPeek, 2011).

7

Analogous to the case for saccades (McPeek & Keller, 2004), the deficits were present only

when the selection was under competitive conditions, i.e., in the presence of distractors. The

strikingly similar role played by the SC in saccade and reach target selection makes a strong case

for a general, effector-independent representation of target “priority” in the SC.

Further support for this claim comes from experiments in freely moving rats performing an odor

discrimination task (Felsen & Mainen, 2008).The rats were trained to turn and walk towards one

of two reward ports on either side of an odor port depending on the identity of an odor or odor

mixture that was presented. SC activity was selective for direction (leftward or rightward) during

locomotion. In many neurons, the selectivity emerged in advance of movement onset, indicating

an active role in selection of the movement. Consistent with the pattern of activity, unilateral

inactivation of SC also produced a profound movement bias towards the ipsiversive location.

Some of the deficits caused by inactivation of the SC seem to share a signature similar to spatial

neglect, albeit one of a localized nature and strictly under competition from distracting stimuli.

SC has a well-known role in directing spatial attention (Cavanaugh & Wurtz, 2004; Krauzlis et

al., 2004), and attentional deficits are wont to cause selection deficits because of their tight

conceptual linkage.

Mechanisms of target selection

Since gaze can be directed to only one location at any time, the distribution of activity in the SC

corresponding to different stimuli must eventually result in one “selected” locus of activity.

Various mechanisms have been proposed to explain how the competition between multiple

stimuli is resolved and to account for the evolution of the target selection signal. The resolution

could follow a simple winner-take-all approach, in which the locus with the highest activity at a

critical time is selected as the target. Indeed, the results of many studies seem to fall under this

category. On the other hand, if the SC is thought to contribute a population code to selection of

the target and the corresponding movement, an appropriately weighted vector averaging or

summation mechanism could be in place (Goossens & Van Opstal, 2006; Katnani, van Opstal, &

Gandhi, 2012; Lee, Rohrer, & Sparks, 1988; Van Opstal & Van Gisbergen, 1990). This approach

is also used under reflexive conditions such as express saccades (Chou, Sommer, & Schiller,

1999). Another possibility is that SC activity at a given locus represents the likelihood of

selecting that location (Kim & Basso, 2010). The distribution of activity across the population

can then be read out by a decoder using Bayesian computations. Using the population activity

recorded at four sites corresponding to the stimulus locations in a visual search task, Kim and

Basso (2010) explicitly tested the predictions of each of these models on a trial-by-trial basis.

The Bayesian maximum a posteriori (MAP) estimate best predicted the saccade selected on a

given trial based on the neural activity at the four locations, outperforming population vector

averaging and winner-take-all.

8

How does the distribution of activity on the SC map emerge in the first place? The canonical

model relies on a lateral recurrent network with local excitatory projections and long-range

inhibitory projections, similar to the organization in primary sensory cortices. Although the basic

physiological properties of SC neurons are consistent with this model, recent findings highlight

the lack of anatomical evidence for such connectivity in the collicular network (Isa & Hall,

2009). The putative function of recurrent lateral inhibition could be served instead by

extracollicular projections, such as from the nuclei isthmii - in birds - and its mammalian

homologues. The cholinergic nucleus isthmi pars parvocellularis (Ipc, cf. parabrachial and

parabigeminal nuclei in mammals) has reciprocal topographic connections with the OT (SC) and

may contribute to feedback amplification of target-related activity. Analogously to the OT, the

Ipc is also known to exhibit switch-like behavior to represent the location of the most salient

stimulus (Asadollahi, Mysore, & Knudsen, 2011). The GABA-ergic nucleus isthmi pars

magnocellularis (Imc, cf. lateral tegmental nucleus in mammals) receives input from the OT and

has anti-topographic projections to the Ipc and OT. This circuit is thought to provide inhibition

that enhances discrimination between target and distractor-related activity (Mysore & Knudsen,

2011b). Inputs from the basal ganglia, notably global GABA-ergic inhibition by the SNr, may

also be a part of the network that facilitates target selection (Hikosaka, Takikawa, & Kawagoe,

2000; Isa & Hall, 2009). Recently, an inhibition-of-inhibition circuit motif has been proposed as

a general mechanism for flexible stimulus categorization based on relative stimulus salience

(Mysore & Knudsen, 2012). Inhibitory interneurons in the SC could be an important part of a

network that implements this computational strategy. Further experiments and models are needed

to identify and delineate the precise role of connections within and outside the SC in target

selection.

Fixation as continuous target selection

As mentioned in the introduction, the visual system is faced with the problem of selecting a

target for most of an animal’s waking life. This problem can be divided into two steps (not

necessarily sequential) - 1) whether to de-select the currently foveated target and select another

target, and, 2) if the previous decision is in the affirmative, which target to select from amongst

the menu of options available to the system. We have already discussed the neural correlates of

the latter in the previous sections. The two can be thought as belonging to the same class of

target selection problems if the following conceptual leap is made: on a moment-by-moment

basis the brain implements (or is implementing) a decision to either continue to “select” the

current stimulus or to select a new target. With this insight, we can now interpret activity in the

SC related to fixation within the target selection framework.

What are the neural correlates of un-fixating and shifting gaze in the SC? Early studies reported

the existence of neurons in the rostral pole of SCid that seemed to mediate gaze-withholding or

the maintenance of fixation (Munoz & Wurtz, 1993). These so-called fixation neurons fire at a

9

tonic rate when the eyes are fixated on a spot and lower their activity during saccades, when

neurons in the caudal SCid show elevated firing. Microstimulation of the rostral SCid stops

saccades in mid-flight (Munoz, Waitzman, & Wurtz, 1996), and can also suppress buildup

activity in the caudal region (Munoz & Istvan, 1998). These observations gave rise to the

fixation zone-saccade zone model, where long-range reciprocal inhibition between the rostral

and caudal “zones” control the alternating pattern of fixations and saccades seen during typical

oculomotor behavior (Dorris, Paré, & Munoz, 1997).

However, numerous lines of evidence have since emerged that dispute this hypothesis. First, the

nature of perturbations in saccades caused by rostral SCid microstimulation fall along a

continuum along with those caused by stimulation of the caudal SCid, and are qualitatively

different from interruptions caused by stimulation of the omnipause neurons (OPNs) in the PPRF

(Gandhi & Keller, 1999). The OPNs exhibit activity that is much more closely linked to the onset

and offset of fixation and therefore make a more qualified candidate for withholding gaze.

Second, experiments in the cat employing multi-step gaze shifts have shown that the locus of

activity across SCid is correlated with the distance to the eventual goal of the sequence of gaze

shifts - or goal-based long term motor error - with rostral SCid neurons returning to their tonic

firing only after the final target is fixated (and not during intermediate fixations) (Bergeron,

Matsuo, & Guitton, 2003). Third, recent work has demonstrated a causal role for rostral SC

neurons in the generation of microsaccades - tiny movements (<1’) of the eyes during fixation

(Hafed, Goffart, & Krauzlis, 2009). These findings are in line with the idea that the SC map

represents a natural continuum of movements, with large movements represented towards the

caudal end and small movements represented near the rostral end. Moreover, the rostro-caudal

distribution of activity may serve as an evolving population code for selecting the final target (or

“goal”) of a gaze shift, ignoring intermediate states that are purely motoric in nature (Krauzlis et

al., 2004). This important distinction is unappreciated in purely single step behavior.

Thus, the transition from withholding gaze to shifting gaze and vice versa is better explained by

a model that considers the shifts of balance between the rostral portion of SCid encoding small

movements (or target errors) and the caudal portion encoding large movements (...). In other

words, reduction of activity in the rostral neurons along with a synchronous rise in the caudal

neurons signifies the de-selection of the currently fixated target and the associated micro-scale

movements while selecting the location represented by the locus of activity in the caudal

population as the target of the next fixation.

Does the time of de-selection of the foveal target necessarily coincide with an impending

movement to a new target or can it be temporally dissociated from the actual saccade? This can

be tested in a modified delayed saccade paradigm, in which the subject is presented a cue during

early fixation to indicate an upcoming presentation of a peripheral target. Recently (unpublished

work), we induced blinks during the initial fixation period in monkeys performing the delayed

10

Figure X.1 — Blink-induced early de-selection of fixated stimulus. (A) The timeline of the task is shown in the

header rows. First row: Activity of a representative neuron in rostral SC on control (black) and blink (gray) trials.

The animal performed in the delayed saccade task, and the blink was evoked an air-puff delivered during fixation,

before the peripheral stimulus was illuminated. The activity is aligned on onset of the blink prior to peripheral target

onset (left), target onset (middle), and onset of saccade (right). For illustration purposes, the control trace in the left

panel (no blink to align on) is a shifted version of the trace from the middle panel. Second row: Eye position traces

aligned on the same three events for blink and control trials. The blink is accompanied by a characteristic loopy

movement that brings the eyes back to the starting point. All eye movements have reached conclusion by the time of

target presentation. (B) Activity of a representative visuo-movement neuron in caudal SC aligned on target onset

(left) and saccade onset (right). Task is the same as in A. Note the increased activity during the visual response and

pre-saccadic build-up on blink trials.

saccade task, prior to the presentation of the eventual saccade target (Gandhi & Jagadisan,

2011). The animals were trained to withhold saccades during the delay period even in the

presence of a peripheral target, and were allowed to look at the target once the fixation spot was

extinguished (GO cue). As shown in Figure X.1A, the activity of rostral SC neurons showed a

sustained reduction following the blink, even when the eyes remained stable and no gaze shift

was produced during this period. Moreover, the reduction of activity in rostral neurons allowed

for increased activation in neurons that had the peripheral target appear in their receptive field

11

(Figure X.1B). This result highlights a putative mechanism of foveal target de-selection in

advance of peripheral target selection and subsequent gaze shift.

It remains to be seen whether the blink plays a special role in forcing the de-selection of the

fixated stimulus or if any associative cue can produce the same effect. It is well-known that the

activity of a sub-population of rostral SC neurons is driven by the presence of a visual stimulus

at the site of fixation; these neurons have lower activity when fixating in the dark (Munoz &

Wurtz, 1993). Since an eye blink transiently occludes the fixated stimulus, it is possible that the

activity of these neurons is reduced due to the lack of visual input, and subsequent, possibly top-

down, mechanisms maintain the activity stable at that level.

Target selection and motor preparation

As discussed above neurons in the intermediate, but not superficial, layers distinguish the

oddball target in a visual search paradigm as early as the second burst of the biphasic visual

response. When studied with clever behavioral tasks designed to probe other processes like

decision-making, confidence evaluation, spatial attention, and reward anticipation, the same class

of SCid neurons also exhibits differential modulation, asserting a contribution of multiple

cognitive mechanisms. Of course, the typical SCid neuron discharges a high-frequency burst

prior to a saccade in its movement field, and it also projects to the brainstem burst generator

elements that execute saccades. Thus, studies have proposed a preparatory premotor component

to the low-level response (Dorris et al., 1997; Glimcher & Sparks, 1992; Hanes & Schall, 1996;

Mazzoni, Bracewell, Barash, & Andersen, 1996; Steinmetz & Moore, 2010) and results

assigning a cognitive role to the neural discharge have been subject to the criticism that the low-

level activity may instead reflect a preparatory command for a movement (usually a saccade) that

is planned but not necessarily executed (Gandhi & Sparks, 2004; Ignashchenkova, Dicke,

Haarmeier, & Thier, 2004; Krauzlis et al., 2004). Of course, it is also likely that both cognitive

and premotor signals are reflected in the low-level discharge, a hypothesis that is the basis of the

“premotor theory of attention” (Rizzolatti, Riggio, Dascola, & Umilta, 1987). While links

between visual attention and saccade preparation have been implied by psychophysical

(Hoffman & Subramaniam, 1995; Rizzolatti et al., 1987) and neurophysiological (Awh,

Armstrong, & Moore, 2006; Cavanaugh & Wurtz, 2004; Corbetta et al., 1998; McPeek & Keller,

2002; Moore & Fallah, 2001; Muller, Philiastides, & Newsome, 2005) studies, evidence also

exists for distinct attention and preparation processes, particularly at the level of frontal eye

fields (Gregoriou, Gotts, & Desimone, 2012; Juan, Shorter-Jacobi, & Schall, 2004b; Schall,

2004; Thompson, Biscoe, & Sato, 2005).

The motor preparation hypothesis states that the low-level discharge in SCid neurons

accumulates gradually toward a cell-specific threshold, at which point (a) it converts into a high-

frequency burst, (b) the brainstem OPNs become quiescent, and (c) a saccade is triggered (Dorris

12

et al., 1997; Hanes & Schall, 1996). It has been hypothesized that the low frequency discharge

represents a motor preparation signal that encodes both timing and metrics of the desired

saccade. Indeed, the firing rate level in the preparatory period is negatively correlated with the

saccade reaction time (the higher the activity, the earlier the movement occurs), and the locus of

activity in the SC dictates the saccade vector. Basic computational models that simulate the

accumulation or drift rate as either noisy (Lo & Wang, 2006; Ratcliff et al., 2003) or ballistic

(Carpenter & Williams, 1995; Reddi et al., 2003) can sufficiently describe the trial-to-trial

variability in the discharge patterns of SCid neurons and the distribution of reaction times. This

functional assessment of motor preparation, however, is gauged by correlating neural activity

with movement features that are observed hundreds of milliseconds later, after the animal is

granted permission to generate a response. A stronger foundation for motor preparation, and its

time course, could be established if a behavioral output can be revealed as the low-frequency

activity is evolving.

Prior to triggering the movement, the saccade generation circuitry must overcome the potent

inhibition imposed to preserve fixation. A potential method to reveal the presence and evolution

of motor preparation exploits the antagonistic relationship between motor preparation and

saccade inhibition. Located in the pons, the OPNs inhibit the saccade burst generator circuit that

innervates the extraocular motoneurons. The OPNs discharge at a tonic rate during fixation and

cease activity during all saccades. Consider the scenario in which the eyes are stable so that the

OPNs are active at a tonic rate, and an experimental manipulation is available to transiently

inhibit them at different times after object(s) are presented in the visual periphery. This early

and transient withdrawal of inhibition could, in principal, “trick” the saccadic system into

prematurely executing an eye movement, if the underlying low-frequency discharge in SCid and

the brainstem burst generator reflects a premotor signal; note that this reasoning remains agnostic

about cognitive signals present simultaneously. If successful, the timing of the earliest saccade

will indicate when motor preparation commences, and the kinematics will reveal how it

develops. Moreover, concurrently recording activity in an oculomotor structure like the SC in

monkeys should provide a neural correlate of timing, kinematics, and direction of prematurely

triggered saccades.

To test this hypothesis, Gandhi and Bonadonna (2005a) utilized the observation that OPNs also

become quiescent during blinks (Schultz, Williams, & Busettini, 2010). An actual connection

between blinks and OPNs may not exist because the pause appears to be associated with the

small, loopy blink-associated eye movement (Schultz et al., 2010). What is important is that

blinks offer a means to remove OPN inhibition. Thus, they delivered an air-puff to one eye to

invoke the trigeminal blink reflex in non-human primates performing various saccade tasks

(Gandhi & Bonadonna, 2005a). Figure X.2A plots temporal traces of eye and eyelid positions of

four reflexive saccades with blinks (color traces) and of an averaged non-blink trial (dashed-

black trace). A blink evoked shortly after target onset but before the typical saccade reaction

13

time triggers a saccade of shorter latency. Figure X.2B illustrates the robust correlation between

saccade latency and blink time across many blink trials. It particularly highlights the time course

of motor preparation, with the shortest reaction times reflecting express saccade latencies.

Figure X.2 -- Behavioral evaluation of the relationship between target selection and motor preparation. (A) Temporal

waveforms of eye and eyelid amplitude as a monkey performed visually guided saccades (bottom part of the panel).

The dashed traces represent a control movement (no blink). The four solid traces denote individual trials in which a

blink was evoked by an air-puff to one eye. The blink (downward deflection in eyelid trace) and the associated

combined blink-saccade waveform can be paired by matching the shades of gray. It can be appreciated that blinks

triggered shortly after the presentation of the peripheral target (and saccade initiation cue) but before a typical saccade

reaction time resulted in a prematurely triggered movement. (B) Saccade latency is plotted as a function of blink time

relative to saccade cue (time zero in panel A) for many trials, each denoted by one symbol (circles and squares identify

leftward and rightward saccades, respectively). The figure highlights the systematic reduction in saccade latency with

blink time. (C) Animals performed a visual search task (one singleton, three distractors), in which the color of the

singleton indicated whether the correct response was a pro- or anti-saccade. Saccade latency is plotted as a function of

blink time for only the subset of trials for which an antisaccade was the correct response. The darkest circles identify

trials with a correct response to the opposite distractor. The lightest circles represent trials with an incorrect response

to an orthogonal distractor. The circles in medium grayscale mark the trials with an incorrect response to the

singleton. (D) The direction of the saccade (0º, singleton; 180º, opposite distractor) is plotted as a function of saccade

latency for the same data shown in panel C. The solid trace represents a moving average of the data. These data

indicate that blink-triggered movements of the shortest latency were directed to the singleton even though the correct

response was to the opposite distractor (180º). Parts of the figure have been adapted from (Gandhi & Katnani, 2011a)

and (Gandhi & Bonadonna, 2005b).

14

The blink perturbation can also be used to probe the time course of motor preparation with

paradigms geared to understand target selection. In the standard visual search task, the singleton

stimulus also serves as the saccade target. Knowledge of the task structure could lead to parallel

processing of target selection and motor preparation, but properties of dissociation or

simultaneity in their time-course would be better appreciated in a task designed to divorce the

singleton location and the required movement vector. In a preliminary study (Gandhi & Katnani,

2011a), we implemented a visual search task that closely resembles that used by Schall and

colleagues (Juan et al., 2004a; Sato & Schall, 2003). Each trial began with several hundred

milliseconds of fixation on a central visual target, which was extinguished when four stimuli

were presented spaced apart by 90º. One target was bestowed a unique feature (color) that

makes it “pop out” from the other, identical “distractors” (McPeek & Keller, 2002; Shen & Paré,

2007; Thompson et al., 1996). The color of the singleton indicated whether the correct response

is a prosaccade to it or an antisaccade to the distractor separated by 180º. Blinks were evoked at

a random time on a small percentage of the trials. Figure X.2C plots the relationship between

blink time and saccade latency for the subset of trials requiring the generation of an antisaccade.

The different symbols identify saccades that were directed incorrectly to the singleton (medium

gray circles), incorrectly to an orthogonal distractor (light gray circles) or correctly to the

opposite location (black circles). When the saccade direction of each movement is plotted as a

function of its latency (Figure X.2D), it becomes clear that saccades with the shortest latencies,

comparable to the time when spatial attention is allocated to the singleton, were incorrectly

driven to the odd-ball stimulus. The transition to the correct location separated by 180º is

observed as the reaction time increases. The blink-perturbation method therefore demonstrates

that motor preparation signal evolves as early as the neural modulation associated with target

selection.

The frontal eye fields have been studied exhaustively for correlates of target selection and motor

preparation in a paradigm that requires antisaccade generation in the context of a visual search

task. The current perspective is spatial attention and motor preparation are dissociated within this

cortical node, particularly within visual neurons. It remains to be determined whether the two

processes can be simultaneously reflected in visuomotor neurons that exhibit both

visual/cognitive selectivity as well as premotor activity. A comparable study has not been

conducted on superior colliculus neurons, yet it seems prudent to search for such signals in this

subcortical structure given its preference for motor-related discharge within the oculomotor

neuraxis (Wurtz, Sommer, Paré, & Ferraina, 2001).

A priority map in the SC

We have established that the SC contains signals that can be used to select a target from non-

targets and, indeed, that the structure plays a causal role in target selection. The studies discussed

in this chapter suggest that the function of SC extends beyond its established role in controlling

15

gaze and gaze-related target selection, towards a more general role in selecting stimuli important

for trans-modal action. Is there a common principle to be gleaned from the activity pattern on the

SC map across experimental paradigms and task requirements?

The usage of the term ‘target’ implicitly refers to the fact that a stimulus is the target of a future

motor act. This definition does not make any assumptions about the nature of the stimulus itself,

beyond its utility for action. However, individual stimuli in the environment are not

homogeneous in their intrinsic properties - some are more luminant than others, some present

better contrast against their surrounding stimuli, and they come in myriad colors. A simplified,

topographic representation of the visual world in terms of the relative strengths of various stimuli

has been proposed, as a salience (or saliency) map (Koch & Ullman, 1985). This representation

is considered bottom-up, independent of the internal state of the animal. On the other hand, the

internal state, including top-down factors such as expectations and goals, can be used to assign

significance to different objects and locations in the visual field, creating a relevance map. An

appropriately weighted, combined representation of bottom-up salience and top-down relevance

produces a priority map, high values on the map indicating high priority for action to the

corresponding location (such as an eye movement or reach). Such maps have been proposed in

various brain regions, including the parietal lobe, frontal eye fields, prefrontal cortex, basal

ganglia, and indeed, the SC (Fecteau & Munoz, 2006). Since the SC is not known to play a direct

role in movements other than gaze shifts (stimulation does not induce other types of movements,

but see Werner, Dannenberg, & Hoffmann, 1997), it has been hypothesized that a generalized,

modality-independent priority map influences action execution in other domains by relaying the

priority information back to the cortex through feedback loops, where it can be parsed into the

appropriate command for effector-specific movements (Song et al., 2011).

If the SCid does in fact house a priority map, it is natural to assume that it does so at all times,

rather than switching to priority map mode under specific circumstances such as visual search or

when competing stimuli are present. Therefore, it is useful to interpret SC activity in simple tasks

such as the single step task within the framework of a priority map. As seen earlier, during

fixation, the population activity in SCid is concentrated towards the rostral end. This can be seen

as a priority map with a single locus in the parafoveal region. Following initial processing of a

stimulus that appears in a peripheral location, the population activity, and thus the priority map,

shifts to that location (Figure X.3). During the gap period in a gap task, the priority map

smoothly shifts from the foveal to a peripheral location, and this is reflected in the inversely

correlated pattern of low frequency build-up activity in rostral and caudal neurons. Thus, the map

evolves as a putative “priority function” of each location in visual space changes in time. The

distribution of activity across the SC population during multi-step gaze shifts is consistent with

an interpretation of this nature. The evolution of the priority signal can also be likened to an

emerging oculomotor “decision” signal observed in perceptual discrimination tasks (Gold &

Shadlen, 2000).

16

Figure X. 3 – Evolution of priority in SC over time. Top row: Delayed visual search task. Only two peripheral

stimuli are used for simplicity. One of the two stimuli is selected as the target based on some feature (not shown).

Gaze location and saccade are shown by dotted lines. Bottom row: SC population activity during the task. During

fixation, the mound of activity is centered around the rostral poles of the two colliculi. At this time, the highest

priority location is the central target. The onset of the search stimuli introduces two new loci of activity on the SC

map. There is a transient shift in priority to these locations around the time of the visual response (time course not

shown), but the priority of the foveal location is maintained higher than the peripheral targets during the delay

period. After permission to saccade, the activity shifts in favor of the selected target (represented in the contralateral

SC) and dies down in the other parts of the map.

The premotor theory of attention can also be tied to the notion of a priority map. One

consequence of viewing ongoing activity in SCid as a priority map is that the motor machinery in

a subpopulation of SCid neurons (motor neurons) and downstream structures has concurrent

access to information about the location with highest priority. Does this imply a direct

relationship between forming a priority map and motor preparation, or are they serial stages in

guiding behavior? In other words, does the priority map map directly onto a space of possible

actions? The evidence presented in the previous section suggests that there may indeed be some

overlap between the two processes.

How are the individual salience and relevance components communicated to the priority map in

the SC? One possibility is that the information about stimulus salience arrives directly or via

other cortical areas from the visual cortex, which is thought to contain a representation of

bottom-up salience and has direct projections to the SC. Recent findings also implicate the

isthmic nuclei (or parabigeminal nuclei) in the classification of stimuli based on their relative

salience. Information about stimulus relevance is expected to arrive from top-down sources

including the respective association cortices and prefrontal cortex. Finally, part of the priority

information could be directly relayed from other cortical areas that are postulated to contain

modality-specific as well as independent representations of priority, including LIP, FEF and

dlPFC.

17

In a recent midbrain-centric account of consciousness (Merker, 2007), the SC is proposed to play

a critical role in a ‘selection triangle’ that involves target selection, action selection, and internal

motivational state. The placement of the colliculus in the command structure of the brain is a key

factor behind the proposal. Of course, it is likely that the SC is part of an extended of network of

regions involved in the selection of targets and actions, albeit with some degree of autonomy

considering its diverse functional properties. The idea of a continuously evolving priority map in

the SC, together with associated maps in the cortex, basal ganglia, and other regions in the

midbrain, fits in neatly within this picture.

Acknowledgments

This work was supported by NIH grant EY015485 and DC0025205.

18

References

Asadollahi, A., Mysore, S. P., & Knudsen, E. I. (2011). Rules of competitive stimulus selection in a cholinergic isthmic nucleus of the owl midbrain. The Journal of neuroscience : the official journal of the Society for Neuroscience, 31(16), 6088-6097.

Awh, E., Armstrong, K. M., & Moore, T. (2006). Visual and oculomotor selection: links, causes and implications for spatial attention. Trends in Cognitive Sciences, 10(3), 124-130.

Basso, M. A., & Wurtz, R. H. (1998). Modulation of neuronal activity in superior colliculus by changes in target probability. The Journal of neuroscience : the official journal of the Society for Neuroscience, 18(18), 7519-7534.

Bergeron, A., Matsuo, S., & Guitton, D. (2003). Superior colliculus encodes distance to target, not saccade amplitude, in multi-step gaze shifts. Nature Neuroscience, 6(4), 404-413.

Berman, R. A., & Wurtz, R. H. (2010). Functional identification of a pulvinar path from superior colliculus to cortical area MT. The Journal of neuroscience : the official journal of the Society for Neuroscience, 30(18), 6342-6354.

Bisley, J. W., & Goldberg, M. E. (2010). Attention, intention, and priority in the parietal lobe. Annual Review of Neuroscience, 33, 1-21.

Carello, C. D., & Krauzlis, R. J. (2004). Manipulating intent: evidence for a causal role of the superior colliculus in target selection. Neuron, 43(4), 575-583.

Carpenter, R. H., & Williams, M. L. (1995). Neural computation of log likelihood in control of saccadic eye movements [see comments]. Nature, 377(6544), 59-62.

Cavanaugh, J., & Wurtz, R. H. (2004). Subcortical modulation of attention counters change blindness. The Journal of neuroscience : the official journal of the Society for Neuroscience, 24(50), 11236-11243.

Chou, I. H., Sommer, M. A., & Schiller, P. H. (1999). Express averaging saccades in monkeys. Vision Research, 39(25), 4200-4216.

Clower, D. M., West, R. A., Lynch, J. C., & Strick, P. L. (2001). The inferior parietal lobule is the target of output from the superior colliculus, hippocampus, and cerebellum. The Journal of neuroscience : the official journal of the Society for Neuroscience, 21(16), 6283-6291.

Corbetta, M., Akbudak, E., Conturo, T. E., Snyder, A. Z., Ollinger, J. M., Drury, H. A., Linenweber, M. R., Petersen, S. E., Raichle, M. E., Van Essen, D. C., & Shulman, G. L. (1998). A common network of functional areas for attention and eye movements. Neuron, 21(4), 761-773.

Dorris, M. C., Paré, M., & Munoz, D. P. (1997). Neuronal activity in monkey superior colliculus related to the initiation of saccadic eye movements. The Journal of neuroscience : the official journal of the Society for Neuroscience, 17(21), 8566-8579.

Everling, S., Dorris, M. C., Klein, R. M., & Munoz, D. P. (1999). Role of primate superior colliculus in preparation and execution of anti-saccades and pro-saccades. Journal of Neuroscience, 19(7), 2740-2754.

Everling, S., & Munoz, D. P. (2000). Neuronal correlates for preparatory set associated with pro-saccades and anti-saccades in the primate frontal eye field. Journal of Neuroscience, 20(1), 387-400.

Fecteau, J. H., & Munoz, D. P. (2006). Salience, relevance, and firing: a priority map for target selection. Trends in Cognitive Sciences, 10(8), 382-390.

Felsen, G., & Mainen, Z. F. (2008). Neural substrates of sensory-guided locomotor decisions in the rat superior colliculus. Neuron, 60(1), 137-148.

Gandhi, N. J., & Bonadonna, D. K. (2005a). Temporal interactions of air-puff-evoked blinks and saccadic eye movements: insights into motor preparation. Journal of neurophysiology, 93(3), 1718-1729.

19

Gandhi, N. J., & Bonadonna, D. K. (2005b). Temporal interactions of air-puff evoked blinks and saccadic eye movements: Insights into motor preparation. Journal of neurophysiology, 93(3), 1718-1729.

Gandhi, N. J., & Jagadisan, U. K. (2011). A neural test bed for simulating executive control deficits in saccade generation. Society for Neurosciences Abstract, Program 18.09.

Gandhi, N. J., & Katnani, H. A. (2011a). Interactions of eye and eyelid movements. In S. P. Liversedge, I. D. Gilchrist & S. Everling (Eds.), Oxford Handbook of Eye Movements: OUP.

Gandhi, N. J., & Katnani, H. A. (2011b). Motor functions of the superior colliculus. Annual Review of Neuroscience, 34, 205-231.

Gandhi, N. J., & Keller, E. L. (1999). Comparison of saccades perturbed by stimulation of the rostral superior colliculus, the caudal superior colliculus, and the omnipause neuron region. Journal of neurophysiology, 82(6), 3236-3253.

Gandhi, N. J., & Sparks, D. L. (2004). Changing views of the role of the superior colliculus in the control of gaze. In L. M. Chalupa & J. S. Werner (Eds.), The Visual Neurosciences (Vol. 2, pp. 1449-1465). Boston: MIT Press.

Glimcher, P. W., & Sparks, D. L. (1992). Movement selection in advance of action in the superior colliculus. Nature, 355(6360), 542-545.

Goldberg, M. E., & Wurtz, R. H. (1972). Activity of superior colliculus in behaving monkey. I. Visual receptive fields of single neurons. Journal of Neurophysiology, 35(4), 542-559.

Goossens, H. H., & Van Opstal, A. J. (2006). Dynamic ensemble coding of saccades in the monkey superior colliculus. Journal of neurophysiology, 95(4), 2326-2341.

Gregoriou, G. G., Gotts, S. J., & Desimone, R. (2012). Cell-type-specific synchronization of neural activity in FEF with V4 during attention. Neuron, 73(3), 581-594.

Hafed, Z. M., Goffart, L., & Krauzlis, R. J. (2009). A neural mechanism for microsaccade generation in the primate superior colliculus. Science, 323(5916), 940-943.

Hanes, D. P., & Schall, J. D. (1996). Neural control of voluntary movement initiation. Science, 274(5286), 427-430.

Hikosaka, O., Takikawa, Y., & Kawagoe, R. (2000). Role of the basal ganglia in the control of purposive saccadic eye movements. Physiological Reviews, 80(3), 953-978.

Hoffman, J. E., & Subramaniam, B. (1995). The role of visual attention in saccadic eye movements. Perception & Psychophysics, 57(6), 787-795.

Huerta, M. F., & Harting, J. K. (1984). The mammalian superior colliculus: Studies of its morphology and connections. In H. Vanegas (Ed.), Comparative Neurology of the Optic Tectum (pp. 687-773). New York: Plenum Publishing.

Ignashchenkova, A., Dicke, P. W., Haarmeier, T., & Thier, P. (2004). Neuron-specific contribution of the superior colliculus to overt and covert shifts of attention. Nature Neuroscience, 7(1), 56-64.

Isa, T., & Hall, W. C. (2009). Exploring the superior colliculus in vitro. Journal of neurophysiology, 102(5), 2581-2593.

Juan, C. H., Shorter-Jacobi, S. M., & Schall, J. D. (2004a). Dissociation of spatial attention and saccade preparation. Proceedings of the National Academy of Sciences USA, 101(43), 15541-15544.

Juan, C. H., Shorter-Jacobi, S. M., & Schall, J. D. (2004b). Dissociation of spatial attention and saccade preparation. Proceedings of the National Academy of Sciences of the United States of America, 101(43), 15541-15544.

Katnani, H. A., van Opstal, A. J., & Gandhi, N. J. (2012). A test of spatial temporal decoding mechanisms in the superior colliculus. Journal of neurophysiology.

Kim, B., & Basso, M. A. (2008). Saccade target selection in the superior colliculus: a signal detection theory approach. The Journal of neuroscience : the official journal of the Society for Neuroscience, 28(12), 2991-3007.

20

Kim, B., & Basso, M. A. (2010). A probabilistic strategy for understanding action selection. The Journal of neuroscience : the official journal of the Society for Neuroscience, 30(6), 2340-2355.

Knudsen, E. I. (1982). Auditory and visual maps of space in the optic tectum of the owl. The Journal of neuroscience : the official journal of the Society for Neuroscience, 2(9), 1177-1194.

Koch, C., & Ullman, S. (1985). Shifts in selective visual attention: towards the underlying neural circuitry. Human neurobiology, 4(4), 219-227.

Krauzlis, R. J., & Dill, N. (2002). Neural correlates of target choice for pursuit and saccades in the primate superior colliculus. Neuron, 35(2), 355-363.

Krauzlis, R. J., Liston, D., & Carello, C. D. (2004). Target selection and the superior colliculus: goals, choices and hypotheses. Vision Research, 44(12), 1445-1451.

Lee, C., Rohrer, W. H., & Sparks, D. L. (1988). Population coding of saccadic eye movements by neurons in the superior colliculus. Nature, 332(6162), 357-360.

Lo, C. C., & Wang, X. J. (2006). Cortico-basal ganglia circuit mechanism for a decision threshold in reaction time tasks. Nature Neuroscience, 9(7), 956-963.

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

Mazzoni, P., Bracewell, R. M., Barash, S., & Andersen, R. A. (1996). Motor intention activity in the macaque's lateral intraparietal area. I. Dissociation of motor plan from sensory memory. Journal of neurophysiology, 76(3), 1439-1456.

McPeek, R. M., & Keller, E. L. (2002). Saccade target selection in the superior colliculus during a visual search task. Journal of neurophysiology, 88(4), 2019-2034.

McPeek, R. M., & Keller, E. L. (2004). Deficits in saccade target selection after inactivation of superior colliculus. Nature Neuroscience, 7(7), 757-763.

Merker, B. (2007). Consciousness without a cerebral cortex: a challenge for neuroscience and medicine. The Behavioral and brain sciences, 30(1), 63-81; discussion 81-134.

Mohler, C. W., & Wurtz, R. H. (1976). Organization of monkey superior colliculus: intermediate layer cells discharging before eye movements. Journal of Neurophysiology, 39(4), 722-744.

Moore, T., & Fallah, M. (2001). Control of eye movements and spatial attention. Proceedings of the National Academy of Sciences USA, 98, 1273-1276.

Moschovakis, A. K., Scudder, C. A., & Highstein, S. M. (1996). The microscopic anatomy and physiology of the mammalian saccadic system. Prog Neurobiol, 50(2-3), 133-254.

Muller, J. R., Philiastides, M. G., & Newsome, W. T. (2005). Microstimulation of the superior colliculus focuses attention without moving the eyes. Proceedings of the National Academy of Sciences of the United States of America, 102(3), 524-529.

Munoz, D. P., & Everling, S. (2004). Look away: the anti-saccade task and the voluntary control of eye movement. Nature reviews. Neuroscience, 5(3), 218-228.

Munoz, D. P., & Istvan, P. J. (1998). Lateral inhibitory interactions in the intermediate layers of the monkey superior colliculus. Journal of neurophysiology, 79(3), 1193-1209.

Munoz, D. P., Waitzman, D. M., & Wurtz, R. H. (1996). Activity of neurons in monkey superior colliculus during interrupted saccades. Journal of neurophysiology, 75(6), 2562-2580.

Munoz, D. P., & Wurtz, R. H. (1993). Fixation cells in monkey superior colliculus. I. Characteristics of cell discharge. Journal of neurophysiology, 70(2), 559-575.

Mysore, S. P., Asadollahi, A., & Knudsen, E. I. (2011). Signaling of the strongest stimulus in the owl optic tectum. The Journal of neuroscience : the official journal of the Society for Neuroscience, 31(14), 5186-5196.

Mysore, S. P., & Knudsen, E. I. (2011a). Flexible categorization of relative stimulus strength by the optic tectum. The Journal of neuroscience : the official journal of the Society for Neuroscience, 31(21), 7745-7752.

21

Mysore, S. P., & Knudsen, E. I. (2011b). The role of a midbrain network in competitive stimulus selection. Current Opinion in Neurobiology, 21(4), 653-660.

Mysore, S. P., & Knudsen, E. I. (2012). Reciprocal inhibition of inhibition: a circuit motif for flexible categorization in stimulus selection. Neuron, 73(1), 193-205.

Nummela, S. U., & Krauzlis, R. J. (2011). Superior colliculus inactivation alters the weighted integration of visual stimuli. The Journal of neuroscience : the official journal of the Society for Neuroscience, 31(22), 8059-8066.

Ratcliff, R., Cherian, A., & Segraves, M. (2003). A comparison of macaque behavior and superior colliculus neuronal activity to predictions from models of two-choice decisions. Journal of neurophysiology, 90(3), 1392-1407.

Reddi, B. A., Asrress, K. N., & Carpenter, R. H. (2003). Accuracy, information, and response time in a saccadic decision task. Journal of neurophysiology, 90(5), 3538-3546.

Rizzolatti, G., Riggio, L., Dascola, I., & Umilta, C. (1987). Reorienting attention across the horizontal and vertical meridians: evidence in favor of a premotor theory of attention. Neuropsychologia, 25(1A), 31-40.

Sato, T. R., & Schall, J. D. (2003). Effects of stimulus-response compatibility on neural selection in frontal eye field. Neuron, 38(4), 637-648.

Schall, J. D. (2001). Neural basis of deciding, choosing and acting. Nature reviews. Neuroscience, 2(1), 33-42.

Schall, J. D. (2004). On the role of frontal eye field in guiding attention and saccades. Vision Research, 44(12), 1453-1467.

Schall, J. D., & Hanes, D. P. (1993). Neural basis of saccade target selection in frontal eye field during visual search. Nature, 366(6454), 467-469.

Schultz, K. P., Williams, C. R., & Busettini, C. (2010). Macaque pontine omnipause neurons play no direct role in the generation of eye blinks. Journal of neurophysiology, 103(4), 2255-2274.

Shen, K., & Paré, M. (2007). Neuronal activity in superior colliculus signals both stimulus identity and saccade goals during visual conjunction search. Journal of Vision, 7(5), 15 11-13.

Sommer, M. A., & Wurtz, R. H. (2004). What the brain stem tells the frontal cortex. I. Oculomotor signals sent from superior colliculus to frontal eye field via mediodorsal thalamus. Journal of neurophysiology, 91(3), 1381-1402.

Song, J. H., Rafal, R. D., & McPeek, R. M. (2011). Deficits in reach target selection during inactivation of the midbrain superior colliculus. Proceedings of the National Academy of Sciences of the United States of America, 108(51), E1433-1440.

Sparks, D. L., & Hartwich-Young, R. (1989). The deep layers of the superior colliculus. Rev Oculomot Res, 3, 213-255.

Steinmetz, N. A., & Moore, T. (2010). Changes in the response rate and response variability of area V4 neurons during the preparation of saccadic eye movements. Journal of Neurophysiology, 103(3), 1171-1178.

Thomas, N. W., & Paré, M. (2007). Temporal processing of saccade targets in parietal cortex area LIP during visual search. Journal of neurophysiology, 97(1), 942-947.

Thompson, K. G., Biscoe, K. L., & Sato, T. R. (2005). Neuronal basis of covert spatial attention in the frontal eye field. The Journal of neuroscience : the official journal of the Society for Neuroscience, 25(41), 9479-9487.

Thompson, K. G., Hanes, D. P., Bichot, N. P., & Schall, J. D. (1996). Perceptual and motor processing stages identified in the activity of macaque frontal eye field neurons during visual search. Journal of neurophysiology, 76(6), 4040-4055.

Van Opstal, A. J., & Van Gisbergen, J. A. (1990). Role of monkey superior colliculus in saccade averaging. Experimental Brain Research, 79(1), 143-149.

Werner, W., Dannenberg, S., & Hoffmann, K. P. (1997). Arm-movement-related neurons in the primate superior colliculus and underlying reticular formation: comparison of neuronal

22

activity with EMGs of muscles of the shoulder, arm and trunk during reaching. Experimental Brain Research, 115(2), 191-205.

Wurtz, R. H., Sommer, M. A., Paré, M., & Ferraina, S. (2001). Signal transformations from cerebral cortex to superior colliculus for the generation of saccades. Vision Research, 41(25-26), 3399-3412.