geniculate nucleus

geniculate nucleus

Eur J Neurosci. 2018;48:3171–3185. wileyonlinelibrary.com/journal/ejn | 3171© 2018 Federation of European Neuroscience Societies and John Wiley & Sons Ltd

Received: 24 January 2018 | Revised: 24 July 2018 | Accepted: 27 July 2018 DOI: 10.1111/ejn.14121

R E V I E W A R T I C L E

Neuronal correlates of motion- defined shape perception in primate dorsal and ventral streams

Takashi Handa1,2 | Akichika Mikami1,3

Edited by Dr. Helen Barbas. Reviewed by Georgia Gregoriou and Arash Yazdanbaksh.

All peer review communications can be found with the online version of the article.

Abbreviations: ITC, inferior temporal cortex; KB, kinetic boundary; LGN, lateral geniculate nucleus; LOC, lateral occipital complex; lSTS, lower bank of the anterior superior temporal sulcus; MRI, magnetic resonance imaging; MT, middle temporal area; PPC, posterior parietal cortex; RF, receptive field; SFL, shape from luminance; SFM, shape from motion; STP, superior temporal polysensory area; uSTS, upper bank of the anterior superior temporal sulcus; V1, primary visual cortex.

1Department of Behavioral and Brain Sciences, Primate Research Institute, Kyoto University, Inuyama, Japan 2Department of Behavior and Brain Organization, Center of Advanced European Studies and Research (CAESAR), Bonn, Germany 3Faculty of Nursing and Rehabilitation, Chubu Gakuin University, Seki, Japan

Correspondence Takashi Handa, Department of Behavior and Brain Organization, Center of Advanced European Studies and Research (CAESAR), Bonn, Germany. Email: [email protected]

Abstract Human and non- human primates can readily perceive the shape of objects using visual motion. Classically, shape, and motion are considered to be separately pro- cessed via ventral and dorsal cortical pathways, respectively. However, many lines of anatomical and physiological evidence have indicated that these two pathways are likely to be interconnected at some stage. For motion- defined shape perception, these two pathways should interact with each other because the ventral pathway must uti- lize motion, which the dorsal pathway processes, to extract shape signal. However, it is unknown how interactions between cortical pathways are involved in neural mech- anisms underlying motion- defined shape perception. We review evidence from psy- chophysical, lesion, neuroimaging and physiological research on motion- defined shape perception and then discuss the effects of behavioral demands on neural activ- ity in ventral and dorsal cortical areas. Further, we discuss functions of two candidate sets of levels: early and higher- order cortical areas. The extrastriate area V4 and middle temporal (MT) area, which are reciprocally connected, at the early level are plausible areas for extracting the shape and/or constituent parts of shape from motion cues because neural dynamics are different from those during luminance- defined shape perception. On the other hand, among other higher- order visual areas, the an- terior superior temporal sulcus likely contributes to the processing of cue- invariant shape recognition rather than cue- dependent shape processing. We suggest that shar- ing information about motion and shape between the early visual areas in the dorsal and ventral pathways is dependent on visual cues and behavioral requirements, indi- cating the interplay between the pathways.

K E Y W O R D S dorsal stream, functional interaction, shape perception, ventral stream, visual motion

www.wileyonlinelibrary.com/journal/ejn
http://orcid.org/0000-0003-3956-8077
mailto:[email protected]
3172 | HANDA AND MIKAMI 1 | INTRODUCTION Among mammals, primates heavily rely on vision. The visual systems of most primates have adapted evolutionally to diur- nal activity. For diurnal primates, visual object recognition plays pivotal roles in the judgment of good foods, such as ripe fruits, and in appropriate action selection, such as catching prey or escaping from predators (Barton, 1996, 1998; Kay & Kirk, 2000). Accordingly, shape perception is a fundamen- tal step in the processing of object recognition. Various vi- sual features, including luminance, color, texture, depth, and motion, enable human and non- human primates to perceive the shape of an object. For instance, visual motion cues are critical in detecting animals that camouflage themselves with a similar color and texture to their surroundings while the animals are still. Once they have moved, it becomes easier for observers to recognize them (Curio, 1976; Eckert & Zeil, 2001; Julesz, 1971; Robinson, 1969) (Figure 1a). In humans, relative motion is the most efficient cue for object segmenta- tion from a visual scene (Nawrot, Shannon, & Rizzo, 1996).

How does the primate brain perform shape perception using such motion cues? A classical view of the primate vi- sual system is that shape and motion are processed through

distinct pathways. Visual information is first transmitted from the retina to the cerebral cortex not only through the lateral geniculate nucleus (LGN) in the thalamus (Leventhal, Rodieck, & Dreher, 1981; Perry, Oehler, & Cowey, 1984; Schiller & Logothetis, 1990) but also through the superior colliculus and inferior pulvinar thalamic nucleus (Berman & Wurtz, 2010; Lyon, Nassi, & Callaway, 2010). Visual transmission through the retino- geniculo pathway has been anatomically and physiologically classified into two parallel pathways. The first is called the parvocellular (or color- opponent) pathway, in which the small receptive fields (RFs) of cells exhibit red- green color- opponent re- sponse patterns and the cells convey sustained signals with spatially fine resolution. A small lesion in the LGN par- vocellular layer has been shown to impair the detection/ discrimination of color, texture, and fine patterns. The sec- ond pathway is called the magnocellular (or broad- band) pathway, in which cells have large RFs and convey achro- matic, low spatial resolution, and more transient signals. A small lesion in the LGN magnocellular layer has been shown to impair motion perception (Derrington & Lennie, 1984; Schiller & Logothetis, 1990; Schiller, Logothetis, & Charles, 1990; Shapley & Perry, 1986). Thus, the parvocel- lular and magnocellular layers are capable of sending sig- nals for processing shape/color and motion, respectively. In the cerebral cortex, two visual pathways originating in the primary visual cortex (V1) have also been characterized. The parvocellular and magnocellular pathways are func- tionally correlated to the ventral and dorsal cortical path- ways, which have been considered to compute non- spatial (shape and color) and spatial (motion and depth) visual features, respectively (Ungerleider & Mishkin, 1982; Van Essen & Gallant, 1994). Among early visual cortical areas, the extrastriate area V4 in the ventral pathway and the mid- dle temporal (MT) area in the dorsal pathway have been extensively profiled. The V4 is critical for shape and color vision (Pasupathy, 2015; Roe et al., 2012), whereas the MT area is dedicated for processing visual motion (Born & Bradley, 2005). Among higher visual cortical areas, the ventral pathway terminates in the inferior temporal cortex (ITC) (Connor, Brincat, & Pasupathy, 2007; Tanaka, 1996; Tompa & Sáry, 2010), whereas the dorsal pathway is di- vided into two side streams that are linked to the posterior parietal cortex (PPC) (Goodale & Milner, 1992; Maunsell & Van Essen, 1983) and anterior superior temporal sulcus (Boussaoud, Ungerleider, & Desimone, 1990) (Figure 2).

For motion- defined shape perception, some corti- cal areas must use motion to extract the boundary be- tween the object and the background or shape of the object. The ventral and dorsal pathways seem to not be wholly independent; rather, they potentially inter- act with each other. In the ventral and dorsal cortical areas, neural inputs originating from the parvocellular

F I G U R E 1 A schematic illustration of motion- defined shape perception. (a) Left: A butterfly camouflaged by its surrounding when still. Once it moves, the shape can be detected by the primate visual system. The white arrow indicates the direction of movement of the butterfly. The white dashed line contour indicates the shape of the butterfly. Right: Extended view around a circle in gray in the left panel. The boundary (white dashed line) is visible by the movement of dots (arrows) on the butterfly against the still dots background. (b) In the laboratory, some artificial motion- defined form stimuli have been used. Left: The kinetic boundary (KB), a visible oriented line (dashed line) at the boundary between the counter movements of dots. Right: Shape from motion (SFM). The relative motion between the inside and outside field of an object enables us to see the shape (circle). Gray arrows indicate the direction of the movement of dots

(a)

(b)

| 3173HANDA AND MIKAMI

and magnocellular pathways physiologically and ana- tomically merge (Maunsell, 1992; Nassi & Callaway, 2009). Parvocellular layer inactivation reduces, but not completely eliminates, visual responses in the V4. Magnocellular layer inactivation comparably reduces the firing rate of V4 neurons in response to an oscil- lating white bar by approximately 40%. Thus, both LGN pathways contribute to visual responses in V4 (Ferrera, Nealey, & Maunsell, 1992; Ferrera, Nealey, & Maunsell, 1994). Moreover, non- direction- selective V4 neurons become tuned to the direction of random dot movement after monkeys have adapted to a visual mo- tion stimulus (Tolias, Keliris, Smirnakis, & Logothetis, 2005). Although the visual responses of MT neurons strongly depend on magnocellular contribution, the responsiveness of a few MT neurons reduced follow- ing parvocellular layer inactivation (Maunsell, Nealey, & DePriest, 1990). Rabies virus tracing has provided further evidence of multisynaptic innervations, which are disynaptic connections linking the magnocellu- lar pathway to the V4 and disynaptic connections the linking parvocellular pathway to the MT (Nassi, Lyon, & Callaway, 2006; Ninomiya, Sawamura, Inoue, & Takada, 2011). Taken together, these findings suggest that the ventral and dorsal pathways can receive each other’s information. This raises the question of how and when such information is used for visual perception. One possibility is that functional interactions between the two pathways are required to achieve motion- defined shape perception. Here we summarize the psy- chophysical and single- cell physiological evidence of motion- defined shape perception. We propose that the coordinated activity between V4 and the MT contrib- utes to the processing of motion- defined shape percep- tion, and discuss future studies that can help uncover

the associated neural circuitry using recently developed approaches.

2 | PSYCHOPHYSICAL AND LESION STUDIES: VISUAL MOTION IS THE MOST EFFICIENT CUE FOR SHAPE PERCEPTION AND THE VENTRAL AND DORSAL PATHWAYS ARE BOTH IMPLICATED IN PROCESSING

Visual motion cues are among the most efficient cues for shape perception and segmentation of objects moving from their background (Braddick, 1993; Nawrot et al., 1996). To determine the efficiency of visual attributes for shape per- ception, threshold levels for accurate shape perception were compared among distinct cues, including luminance contrast, motion, color, density, texture, and binocular disparity. The threshold level for motion- defined shape perception was lower than that for shape perception defined by luminance contrast and color (Nawrot et al., 1996). Relative motion dif- ferences, such as differences in speed or direction between an object and its surrounding background, allow the perception of the boundaries, edges, and contours of shapes (Figure 1b). Psychophysical, electrophysiological and neuroimaging stud- ies have utilized artificial motion- defined stimuli, in which displaying computer- generated random dots at an identical dot density and luminance contrast, but with coherent move- ment of the dots. These conditions allow detection of bound- ary, the so- called kinetic boundary (KB), or object contours, the so- called two- (or three- ) dimensional shape (or structure) from motion (SFM) (Figure 1b). In this review, we specifi- cally refer to two- dimensional motion- defined stimuli. These artificial visual stimuli are useful for investigating the extent

F I G U R E 2 The ventral and dorsal visual pathways in the cerebral cortex of macaque monkeys. Left: The ventral pathway (gray) starts from the V1 and goes to ITC, whereas the dorsal pathway (black) goes to the MT and medial superior temporal (MST) area and then separates into the PPC and uSTS. Right: A coronal section of the brain (indicated by the vertical dashed line in the right panel) showing the uSTS and lSTS where neural activity was recorded (Unno et al. 2014; Handa et al. 2017). The black arrowhead indicates the location of microlesions made after the recordings. A scale bar: 1 cm, sts: superior temporal sulcus, lf: lateral fissure, ips: intraparietal sulcus

lSTS/IT

uSTS

PPC

MST V1V1 MTV4

sts

ips

lSTS

uSTS

lf

D

V L M

IT

3174 | HANDA AND MIKAMI to which their physical features influence perception of ob- servers or responsiveness of neurons by controlling the orien- tations of boundaries, shape of objects, dot density, direction, speed, and coherence of moving dots. Humans and mon- keys are quite good at discriminating the orientations of KB and shapes under the SFM condition (Regan, 1989; Regan & Hamstra, 1991; Schiller, 1993; Sáry, Vogels, & Orban, 1994; Nawrot et al., 1996; Unno, Kuno, Inoue, Nagasaka, & Mikami, 2003). Like humans, macaque monkeys can recog- nize shapes under the SFM condition. The effects of changes in the speed and density of moving dots on SFM perception by monkeys are similar to the effects observed in humans (Unno et al., 2003).

There is increasing evidence from lesion studies to suggest that the ventral and dorsal cortical areas are essential for SFM perception. Damage to the ventral or dorsal cortical regions in humans is related to SFM perception deficits (Mercier, Schwartz, Spinelli, Michel, & Blanke, 2017; Schenk & Zihl, 1997). Deficits in SFM recognition (motion- defined letter) have been found in humans with lesions in parietotempo- ral white matter, which corresponds to Brodmann areas 18, 19, 37, 39, 21, and 22. Some patients have also shown loss of ability in detecting motion and discriminate its direction (Regan, Giaschi, Sharpe, & Hong, 1992). Patients suffering from acute brain damage in the ventral occipito- temporal cortex, in proximity to area MT+/V5, or the lateral occipital complex (LOC) have shown severe SFM perception deficits (Blanke et al., 2007), suggesting that the human ventral and dorsal cortical areas contribute to the processing of motion- defined stimulus perception. In the ventral pathway, the ITC of monkeys plays a pivotal role in recognizing objects (re- tention) and learning new objects (Tanaka, 1996; Tompa & Sáry, 2010). ITC lesions impaired the retention of learned shapes defined by either motion or luminance cues, although learning new object in SFM conditions was less impaired than learning new luminance- defined shapes. Learning per- formance in lesioned monkeys did not differ from perfor- mance in the non- lesioned monkeys (Britten, Newsome, & Saunders, 1992). This result suggests that the ITC plays a role in the discrimination of shapes regardless of cues but that the learning of SFM discrimination is processed by another pathway without the contribution of the ITC. On the other hand, ablation of the MT and adjacent areas in the dorsal pathway impaired the performance of SFM discrimination, but not the performance of luminance- defined shape discrim- ination (Marcar & Cowey, 1992). Lesions in the V4, MT, or both areas impaired accuracy in the judgment of the aspect ratio of rectangles defined by motion cues (Schiller, 1993). Taken together, lesion studies have suggested that the ventral pathway could exclusively play a role in luminance- defined shape perception, but not in SFM perception. In other words, some functional interactions between the ventral and dorsal pathways are required for motion- defined shape recognition.

3 | NEURONAL ACTIVATION IN RESPONSE TO MOTION- DEFINED VISUAL STIMULI IN PASSIVE VIEWING CONDITIONS

A fundamental question is how visual cortical neurons in the ventral and dorsal pathways respond to physical elements of motion- defined shapes. Similar to the psychophysical ap- proach, KB and SFM stimuli have been used to investigate the nature of physiological responsiveness in single neurons regarding the orientation of boundary, density, direction, and speed of moving dots or shapes. For example, the KB, which is made visible as a line by opposing directions of moving dots (Figure 1b, left), may have an orientation orthogonal to that of the axis of the direction of movement. In general, as many visual cortical neurons selectively respond to either the orientation or the direction of motion of conventional moving bars or gratings, we can check if neurons selectively respond to the orientation of the KB or to the direction of dot move- ment. For this purpose, it is reasonable to examine single- unit activity while awake monkeys gaze at a fixation spot after various visual stimuli are presented because multiple stimulus characteristics can be rapidly tested in sequence. Electrophysiological recordings under anesthetized condi- tions are also useful because various stimulus elements can be tested while more stable isolation of the units lasts. Thus, monkeys are passively presented visual stimuli under these conditions.

Neurons in some cortical areas selectively respond to the orientation of a boundary defined by relative motion rather than to the direction of motion. In the ventral pathway, neu- rons in the V2, V4, and ITC selectively responded to the same orientations of the KB even when the directions of moving dots have been orthogonally rotated (Marcar, Raiguel, Xiao, & Orban, 2000; Mysore, Vogels, Raiguel, & Orban, 2006; Sáry, Vogels, & Orban, 1995). A subset of neurons selec- tively responded to motion- defined shapes (i.e., SFM), but the shape selectivity tuning was clearer in the ITC than in the V4 (Mysore, Vogels, Raiguel, & Orban, 2008; Mysore et al., 2006; Sáry, Vogels, & Orban, 1993). In the dorsal path- way, V3A neurons showed orientation selectivity for the KB (Zeki, Perry, & Bartels, 2003). However, MT neurons did not tune to the orientation of the boundary; rather, they tuned to the direction of motion (Marcar, Xiao, Raiguel, Maes, & Orban, 1995). These results suggest that ventral cortical neu- rons primarily extract the KBs or contours of objects using motion cues, while dorsal cortical neurons may partially pro- cess such boundary extractions and mainly contribute to mo- tion processing. However, we must account for other aspects of passive visual stimulation. Even if animals are awake, it is unclear that they actually perceived the given stimuli or recognized the orientation of the KB or the shape of SFM.

| 3175HANDA AND MIKAMI Therefore, it is essential to determine the properties of neu- ronal responses just when monkeys actually recognize pre- sented boundaries and shapes.

4 | ACTIVE VISION MODULATES NEURONAL RESPONSES TO MOTION- DEFINED STIMULUS

It is important to note that the response properties of visual cortical neurons are altered by various task demands (Gilbert & Li, 2013), such as visual attention (McAdams & Maunsell, 2000; Motter, 1994; Ogawa & Komatsu, 2004; Reynolds, Pasternak, & Desimone, 2000; Saruwatari, Inoue, & Mikami, 2008; Treue & Martinez- Trujillo, 1999), visual discrimina- tion (Chelazzi, Duncan, Miller, & Desimone, 1998; Ferrera, Rudolph, & Maunsell, 1994; Handa et al., 2008; Sáry, Köteles, Chadaide, Tompa, & Benedek, 2006; Schlack & Albright, 2007; Sheinberg & Logothetis, 1997), and per- ceptual decision (Newsome, Britten, & Movshon, 1989; Britten, Shadlen, Newsome, & Movshon, 1992; Leopold & Logothetis, 1996; Bradley, Chang, & Andersen, 1998; Nielsen, Logothetis, & Rainer, 2006; Kosai, El- Shamayleh, Fyall, & Pasupathy, 2014; Unno, Handa, Nagasaka, Inoue, & Mikami, 2014). Top- down modulation may enhance the processing of behaviorally significant visual stimuli (Blatt, Andersen, & Stoner, 1990; Schall, Morel, King, & Bullier, 1995; Moore & Armstrong, 2003; Buffalo, Fries, Landman, Liang, & Desimone, 2010; Ninomiya, Sawamura, Inoue, & Takada, 2012; Gregoriou, Rossi, Ungerleider, & Desimone, 2014).

Compared to the passive viewing condition, motion- defined stimulus perception can alter neural activity in the

monkey V1. V1 neurons selective for orientation of KB were scarce in anesthetized condition (Marcar et al., 2000). By con- trast, when monkeys were required to detect the rotation of a line, V1 neurons were more responsive to coherently moving dots perceived as a line but less responsive to the incoherent movement of dots not perceived as a line. The orientation selectivity of V1 neurons for motion- defined lines correlated to the perception of the monkeys, suggesting that the V1 en- codes lines defined by coherent motion signals (Peterhans, Heider, & Baumann, 2005). This discrepancy between pas- sive viewing and visual discrimination conditions may arise from differences in behavioral demands or wake states.

Similar to this comparison, we checked if neuronal acti- vation in visual cortical areas during motion- defined shape perception is different from that during passive viewing and if behavioral demand modulates neuronal activity. To this end, monkeys were required to discriminate motion- defined shapes (SFM) and luminance- defined shapes (shape from luminance, SFL) in a delayed matching- to- sample task (Figure 3). This task paradigm enables us to infer whether monkeys recognize shapes (Vogels & Orban, 1990; Unno et al., 2003). Single- unit activity was extracellularly recorded in the ventral and dorsal cortical areas during task perfor- mance (Figure 4c). More than half of the V4 neurons (57%) showed shape- selective responses to SFM (Figure 4a and d), and the proportion of selective neurons in SFM was larger during shape discrimination (Handa, Inoue, & Mikami, 2010) than during passive viewing (approximately 30%) (Mysore et al., 2008). There was a weak decreasing trend in the shape selectivity of V4 neurons when monkeys made an erroneous choice (Handa et al., 2010). In the MT, approximately 40% of neurons showed shape- modulated activity in response to SFM although their neuronal activity was strongly direction

F I G U R E 3 A delayed matching- to- sample task with motion- defined and luminance- defined shapes. Each trial begins with gazing at a fixation point followed by a sample cue presentation. Monkeys are required to retain the sample shape. After the pseudo- random delay period, two shapes, which are a target and a distractor, are presented during continuous gaze fixation. After the fixation point disappears, the target shape, which is the same as the sample, is chosen by gaze shift (white arrow). When the choice is correct, a reward is given. Otherwise, an error alert is given. Left: SFM condition. Right: Shape from luminance (SFL) condition. Gray arrows indicate the direction of moving dots

Choice

Target Time

SFM condition

or

Delay

Sample

Fixation

SFL condition

3176 | HANDA AND MIKAMI

selective (Figure 4b and d). However, to our knowledge, there is no evidence on the responsiveness of MT neurons to SFM under the passive viewing condition although a paper re- ported that MT neurons did not respond selectively to the ori- entation of KB under the passive viewing condition (Marcar et al., 1995). Thus, the extent to which shape discrimination alters the neural activation in the MT compared with that in the passive viewing condition is unknown.

Further, we addressed the question of whether the require- ment of shape discrimination alters shape- modulated activity in the MT. Neuronal responses to identical SFM stimuli were compared between the requirement of shape discrimination and that of motion discrimination. Of 68 MT neurons, 43% and 24% showed shape- modulated responses when the task required discrimination of shape and direction of motion, re- spectively (Handa et al., 2008). Therefore, the requirement of shape recognition may induce more frequently motion- defined shape modulation in the MT.

We also found that some neurons in the upper bank of the anterior superior temporal sulcus (uSTS), which is also called anterior superior polysensory area (STP) (Baylis, Rolls, & Leonard, 1987; Bruce, Desimone, & Gross, 1981;

Oram & Perrett, 1996), and the lower bank of the ante- rior superior temporal sulcus (lSTS), which are areas TEa and IPa in the ITC (Baylis et al., 1987; Boussaoud et al., 1990), showed selective responses to the shape and motion of SFM during shape discrimination (Unno et al., 2014). The presence of SFM shape- selective uSTS neurons is not consistent with the results of another study in which monkeys were required to gaze at a fixation spot while SFM stimuli were presented without the requirement of shape discrimination. In this passive viewing condition, neuronal modulation by shape in the SFM condition was not observed in the STP (i.e., uSTS) (Anderson & Siegel, 1998). This discrepancy may arise from differences in task demands. Taken together, differences in neural activity between passive and active vison indicate that the recog- nition of motion- defined stimuli can modulate neural re- sponsiveness across the dorsal and ventral pathways. This raises the question of whether enhanced neuronal repre- sentation of motion- defined shapes (or orientations) by the perception of observers in the dorsal and ventral pathways is cue- dependent (i.e. motion or other visual features), or a common neural modulation regardless of cues. Next,

F I G U R E 4 Single- unit activity in V4, MT, anterior superior temporal sulcus during shape discrimination in the SFM condition. Shape- modulated neuronal activity of V4 (a) and MT (b) neurons. The raster and histogram array consist of four shapes and two directions of motion in the SFM condition. This plot shows neural responses to a stimulus presented within its RF. Time is aligned at stimulus onset in target period. (c) A schematic illustration of recording sites. (d) Population histogram of responses to preferred (colored) and no- preferred (gray) stimuli in functionally classified group in V4 (top), MT (middle), and uSTS/lSTS (bottom). (a), (b), and (d) are modified from (Handa et al., 2010; Handa et al., 2008 and Handa et al., 2017), respectively. [Colour figure can be viewed at wileyonlinelibrary.com]

65

65

[Hz]

(a)

30

30

(b)

Motion direction Shape &motion direction Shape

(c)

(d)

N or

m al

iz ed

fi rin

g ra

te

0

0.4

0

0.4

0

0.4 (n = 6) (n = 17) (n = 16)

0 0 0

0.6 0.4 0.6

(n = 8) (n = 36) (n = 21)

0 0

0.6 0.4 (n = 26) (n = 45)

MTV4 uSTS lSTS

0.5 s

MT

V4

[Hz]

www.wileyonlinelibrary.com
| 3177HANDA AND MIKAMI we discuss cue- dependence of neuronal representation of shape by comparison with motion- defined shape and luminance- defined shape perception at different levels of the cortical hierarchy.

5 | CANDIDATE AREAS FOR SHARING OF INFORMATION ABOUT MOTION AND SHAPE ON MOTION- DEFINED STIMULI AT EARLY AND HIGHER- ORDER CORTICAL AREAS

We hypothesize that an interplay between the ventral and dorsal pathways must be required for motion- defined shape (i.e., SFM) perception, but not for luminance- defined shape (i.e., SFL) perception. This hypothesis is based on evidence from lesion studies in which lesions in the dorsal or ventral pathways resulted in distinct effects on the discrimination of shapes defined by motion and luminance cues as discussed above (Britten, Newsome, et al., 1992; Marcar & Cowey, 1992). To this end, we examined functional activity pat- terns in early visual areas and at higher- order cortical areas, which are likely to be candidate areas to share information about motion and shape, based on the following rationale (Figure 2).

At the lower level, the V4 and MT are candidate areas because both areas have direct reciprocal innervations by forming an intermediate connectivity pattern that is dif- ferent from forward and feedback laminar projection pat- terns (Maunsell & Van Essen, 1983; Ungerleider, Galkin, Desimone, & Gattass, 2008). These monosynaptic connec- tions can permit sending and/or receiving information about motion and/or shape to process SFM. V4 can receive visual motion signals through direct connections from MT as well as V2 (Felleman, Burkhalter, & Van Essen, 1997; Maunsell & Van Essen, 1983; Nassi et al., 2006; Ninomiya et al., 2011; Ungerleider et al., 2008), and some V4 neurons encode mo- tion signals (Desimone & Schein, 1987; Ferrera, Rudolph, et al., 1994; Handa et al., 2010; Li et al., 2013; Tolias et al., 2005) (Figure 4a and d). If neural activity is altered between the SFM and SFL conditions, the difference in functional activity may be attributed to differential neural mechanisms underlying shape processing relying on cues. Approximately 50%–60% of shape- selective V4 neurons for SFM revealed a similar shape preference to the SFL, suggesting cue- invariant shape selectivity. However, the temporal properties of shape- modulated neural activity differed between the two cues (Handa, Unno, & Mikami, 2017; Handa et al., 2010; Mysore et al., 2006, 2008). The visual response latency of V4 neurons differed between the SFM and SFL conditions. Delay to represent a shape signal in the V4 was longer in the SFL condition (177 ms) than in the SFM condition (123 ms)

(Handa et al., 2017) (Figure 5). These results suggest that the V4 encodes a specific shape and/or constituent parts of shape regardless of cues, but that the underlying process for the ex- traction of shape and/or its constituent part is dependent on cues (motion vs. luminance). In the MT, the proportion of shape- modulated neurons in the SFM condition (40%) was significantly larger than that in the SFL condition (30%) (Handa et al., 2008). Thus, shape modulation in the MT is dependent on cues. The MT may interact with ventral cortical areas (i.e., V4) in the SFM condition, but less so in the SFL condition. Taken together, in the V4 and MT, shape modula- tion differs depending on cues for shape recognition, indicat- ing that distinct neural mechanisms or circuits are implicated in the processing of shape. This result from the analysis of functional activity supports the interpretation of results from lesion studies (Britten, Newsome, et al., 1992; Marcar & Cowey, 1992; Schiller, 1993).

Another candidate area for the processing of shape using motion at the higher- order level is the anterior superior tem- poral sulcus, where single neurons represent shape and mo- tion signals, but functionally distinct neurons are spatially

F I G U R E 5 Comparison of temporal dynamics of shape- modulated activity between SFM and SFL conditions. (a) Comparison of visual response latency between SFM (gray) and SFL (black) conditions. (b) Comparison of time delay from response latency to emergence of shape representation between SFM and SFL conditions. (a) and (b) are modified from (Handa et al., 2017)

(a)

(b)

0

100

Response latency

Time (ms) 200 0

0

100

lSTS

Time from response latency (ms)

Difference in time

200 0 400

V4

lSTS

V4% o

f c el

l %

o f c

el l

SFL SFM

*

n.s.

*

n.s.

3178 | HANDA AND MIKAMI segregated (Baylis et al., 1987; Jastorff, Popivanov, Vogels, Vanduffel, & Orban, 2012) (Figure 2). The uSTS in the dorsal pathway has reciprocal connections with the medial superior temporal area, which is another motion- sensitive area (Boussaoud et al., 1990), and connections with the ITC (including lSTS) (Saleem, Suzuki, Tanaka, & Hashikawa, 2000). In line with these anatomical connections, uSTS neu- rons showed visually selective responses to moving objects and complex objects such as faces and hands (Bruce et al., 1981; Oram & Perrett, 1996). When monkeys discriminated computer- generated rotating shapes, uSTS neurons selec- tively altered their firing rates in response to the direction of rotation and/or shape of objects (Tanaka, Koyama, & Mikami, 2002). Therefore, the uSTS is considered to play a role in the integration of motion and shape information. On the other hand, in the lSTS, which is a subset of the ITC, neurons responded to simple and complex visual objects (Baylis et al., 1987; Kiani, Esteky, Mirpour, & Tanaka, 2007; Mikami, Nakamura, & Kubota, 1994). A functional magnetic resonance imaging (MRI) study revealed a motion- sensitive subregion in the ITC which may correspond to the lSTS. The SFM or KB induced stronger MRI signal in the lSTS than transparent motion (Nelissen, Vanduffel, & Orban, 2006). Therefore, an interaction between the uSTS and lSTS may enable the areas to extract the SFM signal. In terms of re- sponsiveness to SFM, uSTS neurons primarily showed se- lectivity for the direction of motion, whereas lSTS neurons frequently showed selectivity for the shape (Figure 4d). These results are in accordance with the findings described above, although a minor of uSTS and lSTS neurons showed both shape and direction of motion selectivity (Unno et al., 2014). lSTS neurons may be related to shape recognition regardless

of cues. Most lSTS neurons have a cue- invariant shape pref- erence (Unno et al., 2014), as observed in the lateral convex- ity of the ITC (Sáry et al., 1993; Tanaka, Uka, Yoshiyama, Kato, & Fujita, 2001). Unlike V4 neurons, the temporal as- pects of shape modulation in the lSTS were not significantly different between SFM and SFL conditions (Handa et al., 2017) (Figure 5). The shape selectivity of lSTS neurons sig- nificantly correlated with the shape discrimination perfor- mance of monkeys (Figure 6). At the population level, neural responses to preferred shapes decreased when the shape was a distractor, that is, one of two presented shapes, but not a match to a sample shape. The shape selectivity of lSTS neu- rons was enhanced when monkeys correctly chose shapes matched to sample cues (Unno et al., 2014). Thus, these con- stant neuronal properties of lSTS neurons in response to SFM and SFL suggest that the lSTS plays a role in cue- invariant shape recognition rather than in motion- cue- dependent shape processing.

6 | EARLY CORTICAL AREAS LIKELY PROVIDE A KEY FUNCTION TO SHARE THE INFORMATION OF THE DORSAL AND VENTRAL PATHWAYS

We suggest that interactions between the ventral and dor- sal pathways can be carried out among early cortical areas such as V4 and MT, particularly in the processing of motion- defined shape (Figure 7). V4 is the most likely area because the neuronal response preference for visual features is com- plex; it is crucial for processing of shapes or their constituent

F I G U R E 6 Single- unit activity in lSTS correlates to visual discrimination performance. (a) Neural correlates with shape discrimination performance when the dot density of SFM varies. (b) The rastergram of an lSTS neuron in response to the preferred shape when the monkey selected the correct (up) and incorrect (bottom) shapes. (c) Group data of lSTS neurons in comparison of shape preference between correct and wrong choice trials. (a–c) are modified from (Unno et al., 2014)

correct rate (recording exp.) correct rate (behav exp.) selectivity indices

Correct trials

Error trials

Distractor (n = 7) Target (n = 5)

(a) (b)

C or

re ct

ra te

(% )

Dot density (%)

40

20

1 5 10 20 40

0.8

0.6

0.4

0.2

0

60

80

100 1.0

S electivity index

0 0 4020

0

40

20

Pref – NonPref in error trials

P re

f – N

on P

re f

in c

or re

ct tr

ia ls

–20 –20

(c)

lSTS

| 3179HANDA AND MIKAMI

parts and for transmitting signals to the posterior ITC (Connor et al., 2007), even when visual motion cues are essential for the shape perception (i.e., SFM) (Schiller, 1993). A primary role of V4 may be to facilitate the figure- ground segregation of visual scenes (Roe et al., 2012). Thus, under the SFM con- dition, V4 may extract shape or constituent parts by utilizing a motion signal. In V4, however, the mechanism underlying extraction of shape using motion signals seems to be different from the mechanism underlying the processing of luminance- defined shape. The temporal dynamics of shape representa- tion in V4 are different between the SFM and SFL conditions (Handa et al., 2017). One may consider that V4 receives information about motion or luminance to extract shape in- formation via the dorsal or ventral pathways, respectively (Figure 7). Motion information in V4 may be attributed to that in MT because MT is essential for SFM discrimination. Ablation of MT and adjacent areas has been shown to im- pair SFM discrimination, but not luminance- defined shape discrimination (Marcar & Cowey, 1992). In accordance with this possibility, we found that the emergence of motion infor- mation in V4 followed that in MT (Handa et al., 2017).

Another piece of evidence that supports the interaction be- tween pathways is shape- modulated neuronal activity in the MT. Shape modulation may be enhanced by the requirement of shape discrimination by means of motion cues because neural modulation by shapes has been observed when shapes are related to motion signals (Handa et al., 2008; Schlack & Albright, 2007). These lines of evidence indicate that the MT receives shape signals from the ventral pathway.

An alternative neural mechanism underlying the pro- cessing of SFM extraction is that V2 plays a role in some

interactive processing with V4 and MT. V2 consists of three functionally different compartments called ‘thin- stripes’, ‘thick- stripes’, and ‘inter- stripes’, which are implicated in the processing of color, motion, and orientation, respectively (Van Essen & Gallant, 1994). Therefore, V2 is included in both ventral and dorsal pathways. V2 reciprocally connects with V4 and MT (Maunsell & Van Essen, 1983; Ungerleider et al., 2008) and neuronal activation in V2 is influenced by inactivation of V4 and MT. The orientation or direction se- lectivity were sharpened or lost immediately after the phar- macological inactivation of V4 (Jansen- Amorim, Fiorani, & Gattass, 2012) and MT (Jansen- Amorim, Fiorani, & Gattass, 2011). When MT was inactivated by cooling, the neuronal responses to a moving bar in V2 were modulated (decreased in most cases) at the early (Hupé et al., 2001) and late (Hupé et al., 1998) stages of the responses, indicating that this neu- ral modulation was caused by interference of feedback signal from MT (Hupé et al., 1998). Such a feedback motion signal may be used for the processing of motion- defined stimulus. Indeed, V2 neurons selectively responded to the orientation of the KB and the response latency of the KB- orientation- selective neurons was longer than in non- selective neurons. This temporal difference indicates that the KB- orientation- selective response is computed by using feedback inputs from some areas (Marcar et al., 2000). A model study also has suggested that V4 and MT neurons, which feed signals back to V2 neurons, can play a crucial role in determining figure surfaces distinct from the background using motion cues (Layton & Yazdanbakhsh, 2015) (Figure 7).

At higher- order cortical areas, the uSTS and lSTS are un- likely to work on the extraction of shapes or constituent parts

F I G U R E 7 Possible neural mechanisms across dorsal and ventral pathways underlying SFM and SFL perception. Left: Lateral view of macaque cerebral cortex with labeling of cortical regions. Note that superior temporal sulcus is unfolded in this illustration. Right: A schematic illustration of possible neural mechanism underlying shape processing in SFM (top) and SFL (bottom) conditions with major information flow. Black and gray arrows indicate direction of transmission of motion and shape information, respectively. Unarrowed line indicates anatomical connection on the basis of literature: a, Maunsell & Van Essen (1983); b, Ungerleider et al. (2008); c, Boussaoud et al. (1990); d, Saleem et al. (2000). [Colour figure can be viewed at wileyonlinelibrary.com]

SFM

SFL

a,b b

a c

d

a,b b

a c

d

b

b

a

a a

a

SFM

SFL MST uSTS

lSTS ITC

V1 V2

MT

V4

MST uSTS

lSTS ITC

? V1

MT

V4 V2

motion shape/boundary

a,b b

a c

d

a,b b

a c

d

MT

uSTS lSTS

V4 MST

V1 V2

PPC

ITC

b

b

a

a a

a

www.wileyonlinelibrary.com
3180 | HANDA AND MIKAMI (boundaries) using motion cues. Rather, the lSTS encodes in- variant shape information and can contribute to shape recog- nition. Cue- invariant orientation selective neurons in the ITC are less common than in the V4 (Mysore et al., 2006; Sáry et al., 1995), whereas cue- invariant shape- selective neurons are enriched (Sáry et al., 1993; Tanaka et al., 2001; Unno et al., 2014). Shape selectivity for the SFM was correlated to discrimination performance in monkeys (Unno et al., 2014). The temporal dynamics of shape- selective responses in the lSTS did not differ between SFM and SFL conditions (Handa et al., 2017). Taken together, the lSTS and ITC contribute to invariant shape recognition rather than to encoding cue- dependent curvature or shape contours.

In our study, the uSTS may not have contributed to shape processing under the SFM condition because few neurons showed shape selectivity for the SFM. This result may be due to the usage of simple shapes as uSTS neurons are more likely to respond to complex objects, such as faces and hands, and to body movements and biological motion, but not to sim- ple shapes (Baylis et al., 1987; Bruce et al., 1981; Oram & Perrett, 1996). In the uSTS and lSTS, the integration of mo- tion and object information may be required to encode what object is moving where (e.g., a human walking forward). In addition to the previous single- unit electrophysiological find- ings in the uSTS (Bruce et al., 1981; Oram & Perrett, 1996), recent studies demonstrated that various sub- regions in su- perior temporal sulcus, including the lSTS, were activated by observing artificial biological motion (Vangeneugden, Pollick, & Vogels, 2009) and observing action (Nelissen et al., 2011). Human neuroimaging studies indicate that the posterior superior temporal sulcus, which corresponds with the monkey uSTS, is involved in the perception of biological motion, and the function of this cortical area is implicated in the integration/separation of shape and motion (Jastorff & Orban, 2009). In the human ventral pathway, the extrastri- ate body area, which is involved in the analysis of the static human body form, was selectively activated by biological motion displays (Peelen, Wiggett, & Downing, 2006). Thus, interplay between the uSTS and lSTS may contribute to the processing of higher cognitive functions such as action rec- ognition (Keysers & Perrett, 2004).

Functional MRI and event- related potential mapping in human brains have implicated the ventral and dorsal visual pathways in motion- defined shape perception (for review, see Kourtzi, Krekelberg, & Van Wezel, 2008). The dorsal visual area MT+/V5, which corresponds with the monkey MT, as well as ventral and lateral occipital areas such as the LOC and ventral occipitotemporal cortex, which are considered to cor- respond with the monkey ITC, showed SFM- related activa- tion (Gulyas, Heywood, Popplewell, Roland, & Cowey, 1994; Mercier et al., 2017; Schoenfeld et al., 2003; Wang et al., 1999). Neuroimaging studies uncovered the involvement of multiple cortical areas in dorsal and ventral pathways, such

as V3, the fundus of the superior temporal sulcus, and PPC in addition to V2, V4 and MT, in the processing of three- dimensional structure- from- motion (for review, see Orban, 2011). Thus, we should consider that the processing of two- dimensional SFM involves a large scale network that includes dorsal and ventral cortical areas.

7 | CONCLUDING REMARKS AND FUTURE PERSPECTIVES

Considerable physiological and anatomical evidence as well as lesion studies have indicated cooperation between the ven- tral and dorsal cortical areas for visual perception (Maunsell, 1992; Gegenfurtner & Hawken, 1996; Perry & Fallah, 2014), which is inconsistent with the classical concept, namely the separation of the two visual pathways. A recent paper revealed that shape- sensitive fMRI signals in the human ventral and dorsal higher- order regions were correlated to shape perception of observers, suggesting involvement of both pathways in shape perception (Freud, Culham, Plaut, & Behrmann, 2017). Not only the ventral pathway but also the dorsal pathway may contribute to the processing of general object recognition (Farivar, 2009).

The analysis of neuronal activity during shape recognition has provided evidence that neural activation in the cortical areas of the ventral and dorsal pathways is more modulated by SFM discrimination than in the passive viewing condition. The temporal dynamics of neuronal modulation by SFM is different from those of neuronal modulation by luminance- defined shape, suggesting that the underlying neural mech- anisms are distinct for motion and luminance cues. There is increasing anatomical and physiological evidence to support the concept of an interactive mechanism between the MT and V4. Motion- defined shape perception can be computed via the ventral pathways with the cooperation of motion- sensitive cortical areas such the MT. A KB or SFM can be extracted at an early level such as V4. By contrast, higher- order ventral cortical areas, such as the lSTS and ITC, encode invariant shape information and directly contribute to shape recognition, rather than the extraction of kinetic edges or sim- ple shapes defined by motion.

Admittedly, there is so far no direct evidence to support or refute the interactive neural computation between MT and V4 in response to motion- defined shape. To determine the presence of the interactive function at early visual cor- tical areas, it is ideally required to identify neurons anatom- ically connecting to other pathways, to analyze the encoded signal (shape or motion) that the identified neurons convey, and to investigate the effect of the inactivation of the iden- tified neurons on neural representation at the other region. Technical advances in rodent brain research, such as opto- genetic manipulation (transient and reversal depolarization

| 3181HANDA AND MIKAMI or hyperpolarization) and large scale multiple- neuron re- cordings, have enabled us to investigate the involvement of a specific brain region, neural circuit, or pathway in a specific behavior (Bolkan et al., 2017; Saiki et al., 2018). Importantly, optogenetics can identify specific projection neurons and can be used in primate brain research (Klein et al., 2016; Stauffer et al., 2016). A combination of op- togenetic manipulation in specific pathways with electro- physiological recording (Cavanaugh et al., 2012; Inoue, Takada, & Matsumoto, 2015) or functional MRI (Gerits & Vanduffel, 2013) can be useful for understanding com- prehensive networks relevant to specific processing (e.g., motion- defined shape perception) and the causality of per- ception (Kahn et al., 2013). Future studies need to chal- lenge the investigation of interactive functions at more precise connection levels.

ACKNOWLEDGEMENTS

We are grateful to all of our colleagues of primate research institute for their supports.

CONFLICT OF INTEREST

None declared.

AUTHOR CONTRIBUTIONS

AM designed and directed the project. TH performed experi- ments, analyzed the data, and wrote the manuscript.

ORCID

Takashi Handa http://orcid.org/0000-0003-3956-8077

REFERENCES

Anderson, K. C., & Siegel, R. M. (1998). Lack of selectivity for simple shapes defined by motion and luminance in STPa of the behaving macaque. NeuroReport, 9, 2063–2070. https://doi. org/10.1097/00001756-199806220-00028

Barton, R. A. (1996). Neocortex size and behavioural ecology in primates. Proceedings of the Royal Society of London, Series B: Biological Sciences, 263, 173–177. https://doi.org/10.1098/ rspb.1996.0028

Barton, R. A. (1998). Visual specialization and brain evolution in primates. Proceedings of the Royal Society of London, Series B: Biological Sciences, 265, 1933–1937. https://doi.org/10.1098/ rspb.1998.0523

Baylis, G. C., Rolls, E. T., & Leonard, C. M. (1987). Functional subdi- visions of the temporal lobe neocortex. Journal of Neuroscience, 7, 330–342. https://doi.org/10.1523/JNEUROSCI.07-02-00330.1987

Berman, R. A., & Wurtz, R. H. (2010). Functional identification of a pulvinar path from superior colliculus to cortical area MT.

Journal of Neuroscience, 30, 6342–6354. https://doi.org/10.1523/ JNEUROSCI.6176-09.2010

Blanke, O., Brooks, A., Mercier, M., Spinelli, L., Adriani, M., Lavanchy, L., … Landis, T. (2007). Distinct mechanisms of form- from- motion perception in human extrastriate cortex. Neuropsychologia, 45, 644–653. https://doi.org/10.1016/j. neuropsychologia.2006.07.019

Blatt, G. J., Andersen, R. A., & Stoner, G. R. (1990). Visual receptive field organization and cortico- cortical connections of the lateral in- traparietal area (area LIP) in the macaque. Journal of Comparative Neurology, 299, 421–445. https://doi.org/10.1002/(ISSN)1096-9861

Bolkan, S., Stujenske, J. M., Parnaudeau, S., Spellman, T. J., Rauffenbart, C., Abbas, A. I., … Kellendonk, C. (2017). Thalamic projections sustain prefrontal activity during working memory maintenance. Nature Neuroscience, 20, 987–996. https://doi.org/10.1038/nn.4568

Born, R. T., & Bradley, D. C. (2005). Structure and function of visual area MT. Annual Review of Neuroscience, 28, 157–189. https://doi. org/10.1146/annurev.neuro.26.041002.131052

Boussaoud, D., Ungerleider, L. G., & Desimone, R. (1990). Pathways for motion analysis: Cortical connections of the medial superior temporal and fundus of the superior temporal visual areas in the macaque. Journal of Comparative Neurology, 296, 462–495. https:// doi.org/10.1002/(ISSN)1096-9861

Braddick, O. (1993). Segmentation versus integration in visual mo- tion processing. Trends in Neurosciences, 16, 263–268. https://doi. org/10.1016/0166-2236(93)90179-P

Bradley, D. C., Chang, G. C., & Andersen, R. A. (1998). Encoding of three- dimensional structure- from- motion by primate area MT neu- rons. Nature, 392, 714–717. https://doi.org/10.1038/33688

Britten, K. H., Newsome, W. T., & Saunders, R. C. (1992). Effects of inferotemporal cortex lesions on form- from- motion discrimination in monkeys. Experimental Brain Research, 88, 292–302. https://doi. org/10.1007/BF02259104

Britten, K. H., Shadlen, M. N., Newsome, W. T., & Movshon, J. A. (1992). The analysis of visual motion: A comparison of neuronal and psychophysical performance. Journal of Neuroscience, 12, 4745– 4765. https://doi.org/10.1523/JNEUROSCI.12-12-04745.1992

Bruce, C., Desimone, R., & Gross, C. G. (1981). Visual properties of neurons in a polysensory area in superior temporal sulcus of the macaque. Journal of Neurophysiology, 46, 369–384. https://doi. org/10.1152/jn.1981.46.2.369

Buffalo, E. A., Fries, P., Landman, R., Liang, H., & Desimone, R. (2010). A backward progression of attentional effects in the ven- tral stream. Proceedings of the National Academy of Sciences, 107, 361–365.

Cavanaugh, J., Monosov, I. E., McAlonan, K., Berman, R., Smith, M. K., Cao, V., … Wurtz, R. H. (2012). Optogenetic inactivation modi- fies monkey visuomotor behavior. Neuron, 76, 901–907. https://doi. org/10.1016/j.neuron.2012.10.016

Chelazzi, L., Duncan, J., Miller, E. K., & Desimone, R. (1998). Responses of neurons in inferior temporal cortex during memory- guided visual search. Journal of Neurophysiology, 80, 2918–2940. https://doi.org/10.1152/jn.1998.80.6.2918

Connor, C. E., Brincat, S. L., & Pasupathy, A. (2007). Transformation of shape information in the ventral pathway. Current Opinion in Neurobiology, 17, 140–147. https://doi.org/10.1016/j.conb.2007.03.002

Curio, E. (1976). The ethology of predation, Vol. 7. Berlin-Heidelberg-New York: Springer Verlag. https://doi.org/10.1007/978-3-642-81028-2

Derrington, A. M., & Lennie, P. (1984). Spatial and temporal contrast sensitivities of neurones in lateral geniculate nucleus of macaque.

http://orcid.org/0000-0003-3956-8077
http://orcid.org/0000-0003-3956-8077
https://doi.org/10.1097/00001756-199806220-00028
https://doi.org/10.1097/00001756-199806220-00028
https://doi.org/10.1098/rspb.1996.0028
https://doi.org/10.1098/rspb.1996.0028
https://doi.org/10.1098/rspb.1998.0523
https://doi.org/10.1098/rspb.1998.0523
https://doi.org/10.1523/JNEUROSCI.07-02-00330.1987
https://doi.org/10.1523/JNEUROSCI.6176-09.2010
https://doi.org/10.1523/JNEUROSCI.6176-09.2010
https://doi.org/10.1016/j.neuropsychologia.2006.07.019
https://doi.org/10.1016/j.neuropsychologia.2006.07.019
https://doi.org/10.1002/(ISSN)1096-9861
https://doi.org/10.1038/nn.4568
https://doi.org/10.1146/annurev.neuro.26.041002.131052
https://doi.org/10.1146/annurev.neuro.26.041002.131052
https://doi.org/10.1002/(ISSN)1096-9861
https://doi.org/10.1002/(ISSN)1096-9861
https://doi.org/10.1016/0166-2236(93)90179-P
https://doi.org/10.1016/0166-2236(93)90179-P
https://doi.org/10.1038/33688
https://doi.org/10.1007/BF02259104
https://doi.org/10.1007/BF02259104
https://doi.org/10.1523/JNEUROSCI.12-12-04745.1992
https://doi.org/10.1152/jn.1981.46.2.369
https://doi.org/10.1152/jn.1981.46.2.369
https://doi.org/10.1016/j.neuron.2012.10.016
https://doi.org/10.1016/j.neuron.2012.10.016
https://doi.org/10.1152/jn.1998.80.6.2918
https://doi.org/10.1016/j.conb.2007.03.002
https://doi.org/10.1007/978-3-642-81028-2
3182 | HANDA AND MIKAMI Journal of Physiology, 357, 219–240. https://doi.org/10.1113/ jphysiol.1984.sp015498

Desimone, R., & Schein, S. J. (1987). Visual properties of neurons in area V4 of the macaque: sensitivity to stimulus form. Journal of Neurophysiology, 57, 835–868. https://doi.org/10.1152/jn.1987.57.3.835

Eckert, P. M., & Zeil, J. (2001). Toward an ecology of motion vision. In J. M. Zanker, & J. Zeil (Eds.), Motion vision: Computational, neural, and ecological constraints (pp. 333–369). Berlin: Springer. https://doi.org/10.1007/978-3-642-56550-2

Farivar, R. (2009). Dorsal- ventral integration in object recogni- tion. Brain Res. Reviews, 61, 144–153. https://doi.org/10.1016/j. brainresrev.2009.05.006

Felleman, D. J., Burkhalter, A., & Van Essen, D. C. (1997). Cortical connections of area V3 and VP of macaque monkey extrastriate vi- sual cortex. Journal of Comparative Neurology, 379, 21–47. https:// doi.org/10.1002/(ISSN)1096-9861

Ferrera, V. P., Nealey, T. A., & Maunsell, J. H. (1992). Mixed parvocel- lular and magnocellular geniculate signals in visual area V4. Nature, 358, 756–761. https://doi.org/10.1038/358756a0

Ferrera, V. P., Nealey, T. A., & Maunsell, J. H. (1994). Responses in ma- caque visual area V4 following inactivation of the parvocellular and magnocellular LGN pathways. Journal of Neuroscience, 14, 2080– 2088. https://doi.org/10.1523/JNEUROSCI.14-04-02080.1994

Ferrera, V. P., Rudolph, K. K., & Maunsell, J. H. (1994). Responses of neurons in the parietal and temporal visual pathways during a motion task. Journal of Neuroscience, 14, 6171–6186. https://doi. org/10.1523/JNEUROSCI.14-10-06171.1994

Freud, E., Culham, J.C., Plaut, D.C., & Behrmann, M. (2017). The large- scale organization of shape processing in the ventral and dor- sal pathways. eLife, 6, e34464.

Gegenfurtner, K. R., & Hawken, M. J. (1996). Interaction of motion and color in the visual pathways. Trends in Neurosciences, 19, 394–401.

Gerits, A., & Vanduffel, W. (2013). Optogenetics in primates: A shin- ing future? Trends Genetics, 29, 403–411. https://doi.org/10.1016/j. tig.2013.03.004

Gilbert, C. D., & Li, W. (2013). Top- down influences on visual pro- cessing. Nature Reviews Neuroscience, 14, 350–363. https://doi. org/10.1038/nrn3476

Goodale, M. A., & Milner, A. D. (1992). Separate visual pathways for perception and action. Trends in Neurosciences, 15, 20–25. https:// doi.org/10.1016/0166-2236(92)90344-8

Gregoriou, G. G., Rossi, A. F., Ungerleider, L. G., & Desimone, R. (2014). Lesions of prefrontal cortex reduce attentional modulation of neuronal responses and synchrony in V4. Nature Neuroscience, 17, 1003–1011. https://doi.org/10.1038/nn.3742

Gulyas, B., Heywood, C. A., Popplewell, D. A., Roland, P. E., & Cowey, A. (1994). Visual form discrimination from color or motion cues: Functional anatomy by positron emission tomography. Proceedings of the National Academy of Sciences, 91, 9965–9969. https://doi. org/10.1073/pnas.91.21.9965

Handa, T., Inoue, M., & Mikami, A. (2010). Neuronal activity during discrimination of shapes defined by motion in area V4. NeuroReport, 21, 532–536. https://doi.org/10.1097/WNR.0b013e3283393a5f

Handa, T., Katai, S., Kuno, R., Unno, S., Inoue, M., & Mikami, A. (2008). Differential activity to shapes under shape- from- motion condition in macaque middle temporal area. Neuroscience, 156, 1118–1135. https://doi.org/10.1016/j.neuroscience.2008.06.029

Handa, T., Unno, S., & Mikami, A. (2017). Temporal property of single- cell activity in response to motion- defined shapes in monkey dorsal

and ventral cortical areas. NeuroReport, 28, 793–799. https://doi. org/10.