rationalskepticism.org

Are you good with faces? So is the Japanese rice fish – at least, it is if the faces are the right way up. Just like humans, the tiny fish has no problem recognising faces orientated the usual way, but, again like us, it struggles when they are inverted. The finding indicates that the fish may have developed a unique brain pathway for face recognition, just as humans have.

We have no problem identifying most objects in our environment – say, a chair – no matter what way up they are. But faces are different. It is relatively easy for us to spot the differences between two faces, even if they are physically similar, if we see them in photographs the right way up. But if the images are upside down, telling them apart gets a bit tricky.

“This is because we have a specific brain area for processing faces, and when the face is upside down, we process the image through object processing pathways, and not the face-processing pathways any more,” says Mu-Yun Wang at the University of Tokyo, Japan.

Until now, this face-inversion effect was considered exclusive to mammals as it has only been observed in primates and sheep.

Enter the Japanese rice fish, also known as the medaka (Oryzias latipes), a 3.5-centimetre-long shoaling fish commonly found in rice paddies, marshes, ponds and slow-moving streams in East Asia. These fish are very social, so identifying the right individuals to associate with is important.

Wang & Takeuchi, 2017 wrote:Abstract

Individual recognition (IR) is essential for maintaining various social interactions in a group, and face recognition is one of the most specialised cognitive abilities in IR. We used both a mating preference system and an electric shock conditioning experiment to test IR ability in medaka, and found that signals near the face are important. Medaka required more time to discriminate vertically inverted faces, but not horizontally shifted faces or inverted non-face objects. The ability may be comparable to the classic ‘face inversion effect’ in humans and some other mammals. Extra patterns added to the face also did not influence the IR. These findings suggest the possibility that the process of face recognition may differ from that used for other objects. The complex form of recognition may promote specific processing adaptations, although the mechanisms and neurological bases might differ in mammals and medaka. The ability to recognise other individuals is important for shaping animal societies.

Wang & Takeuchi, 2017 wrote:Being able to recognize each other is crucial for social interactions in humans, as well as many other animals. To humans, faces are the most important body part to differentiate between one another. Humans read the face as a whole, rather than look at parts of the face, which is why it is harder to recognise a face when we see it upside-down, but not when we see an upside-down object.

Some other mammals also identify each other by the face and take longer to recognise an upside-down face, but this ability has never been observed in animals other than mammals. Previous research has shown that some fish species can distinguish between individuals. For example, female medaka fish prefer males they have seen before to ‘strangers’. However, until now, it was not known if they can recognize individual faces, nor how they distinguish a specific male from many others.

To see if medaka fish use vision, smell or both cues to recognise mates, Wang and Takeuchi familiarised the fish before the mating test in different settings. In the first group, the male and the female could see each other but were kept in different tanks; in the second group to test odour cues, the male and the female were in the same tank but could not see each other; in the third group, the fish were in the same tank and could see each other; the fish in the fourth group were kept in different tanks and could not see each other. To make sure the fish can recognise and distinguish between fish or objects, Wang and Takeuchi also performed negative conditioning experiments, in which the females had to learn to form an association between a negative stimulus and a specific situation.

Wang and Takeuchi found that medaka fish use both vision and smell to distinguish between other fish, but could recognise each other based on vision alone. More specifically, the fish looked at the faces to tell others apart, and even when spots were added to their faces, the fish could still recognise the other. The mekada fish were also able to discriminate between two fish and two objects, but failed the task when the fish images were presented upside-down. However, when two objects were inverted, they were still able to tell the difference. This suggests that just like humans, faces may be special for fish too.

This is the first study that shows the face inversion effect in animals other than mammals. A next step will be to compare the different mechanisms between species, and identify the underlying genes and nerve cells responsible for face recognition. This will enable us to better understand social interactions in fish, and enhance our knowledge of how our own ability to recognize faces has changed from an evolutionary point of view.

Wang & Takeuchi, 2017 wrote:Introduction

In a social group, the ability to recognise other individuals correctly is essential for maintaining various social interactions in animals, such as pair-bonding, hierarchy, inbreeding avoidance, and recognition of offspring, nest mates, or neighbours (Tibbetts and Dale, 2007; Wiley, 2013). For example, some territorial birds can remember specific neighbours for a long period of time (Godard, 1991), and king penguins can identify their chick from thousands of conspecifics (Aubin and Jouventin, 1998). Receivers associate different types of identity signals, such as odour, sound, tactile, motion, electric or morphological cues, with certain individuals (Sherman et al., 2009) and identify them afterwards when necessary. In addition to looking at how animals recognise conspecifics, their mental representations of specific individuals can also give hints that allow us to judge their cognitive abilities. For example, hamsters have various odours for different body parts, and an unfamiliar hamster will categorise them as multiple individuals, while a previously interacted hamster can associate the odours to the specific individual (Johnston and Bullock, 2001). Animals may have complicated mechanisms to link multiple identity signals to different types of fitness-related tasks, or may use simpler rules to remember an individual. Among all of the individual recognition (IR) systems, face recognition is one of the most specific abilities, and is reported in animals from a number of distinct evolutionary lineages (Kendrick and Baldwin, 1987; McKone et al., 2007; Van der Velden et al., 2008, Coulon et al., 2009; Racca et al., 2010; Sheehan and Tibbetts, 2011). How faces are recognised, and whether the processes involved differ from those used to perceive other objects, is a main topic of interest in the field of cognitive psychology and biology.

In humans and some other mammals, faces are specially processed in cognitive, developmental and functional ways (Calder, 2011). Human infants are hypothesised to be attracted to faces innately (Morton and Johnson, 1991), but also develop face recognition skills and specific brain regions for processing faces during childhood. A familiar face can be individuated in 250 ms (Jacques and Rossion, 2006), and we can possibly remember more faces than other visual stimuli with similar variations in details and features. Studies of a neuropsychological disorder known as prosopagnosia or face blindness, in which individuals are unable to recognise faces but have no difficulty in recognising individuals by other modalities (such as voice) or in discriminating non-face objects (Meadows, 1974; Behrmann et al., 2005), have shown that facial recognition proceeds through specific cognitive and neural pathways (Valentine, 1988; McKone et al., 2007). In addition, the increase in recognition difficulty associated with inversion of faces is greater than that for the inversion of other types of visual stimuli (Yin, 1969). The so-called face-inversion effect indirectly indicates that faces are perceived configurally rather than only by specific features (such as the eyes, nose, or mouth), and that once inverted, such a global configuration is difficult to match and passes through routes which are used for recognising other objects (Bartlett and Searcy, 1993; Haxby et al., 1999; Boutsen et al., 2006). Likewise, the Thatcher illusion found in both humans (Thompson, 1980) and monkeys (Adachi et al., 2009; Dahl et al., 2010), in which the eyes and mouth are inverted relative to the face, becomes difficult to detect when upside down, further demonstrating that configural perception is interrupted when orientation is inverted.

Some other animals, ranging from mammals, birds and fish to invertebrates, have also been reported to use faces for IR (Brown and Dooling, 1992; Kendrick et al., 1995; Bovet and Vauclair, 2000; Van der Velden et al., 2008, Kohda et al., 2015; Parr and Hecht, 2011; Tibbetts, 2002). Scientists have long argued that the face-specific processes are unique to humans or shared only by quite closely related species (Tate et al., 2006). However, such specialised ability may also have evolved in distinct animal taxa when selection force associated with complicated, repeated social interactions strongly favours IR.

The face inversion effect is the method most widely used in animals to test whether faces may be processed specifically, and researchers have identified this ability in some non-human primates (Overman and Doty, 1982; Tomonaga, 1994; Parr et al., 1998; Vermeire and Hamilton, 1998; Neiworth et al., 2007) and in sheep (Kendrick et al., 1996). Some monkeys failed to show such oriented-specific face-processing (Rosenfeld and Van Hoesen, 1979; Bruce, 1982; Dittrich, 1990; Parr et al., 1999; Weiss et al., 2001; Gothard et al., 2004), but many studies lacked the use of non-face signals as controls, making it difficult to interpret the results (Parr et al., 1999). Specialised neural systems for face recognition have been found in some non-human primates and in sheep (Kendrick and Baldwin, 1987; Kanwisher and Yovel, 2006; Tsao et al., 2006), providing great opportunities to interpret how these animals perceive faces perceptually and mechanically for comparative research. Other than the inversion effect, sheep, chimpanzees, and wasps exhibit better discrimination of conspecific faces than of non-face objects (Kendrick et al., 1996; Parr et al., 1998; Sheehan and Tibbetts, 2011). The difference between decision speed and accuracy in discriminating faces and non-facial stimuli is hypothesized to be due to face-specific perception (Sheehan and Tibbetts, 2011).

In the present study, we used a popular freshwater animal model, the medaka fish (Figure 1A), to test IR ability and to examine whether these animals perceive faces differently from non-face stimuli. Researchers have only recently found that fish can use facial pattern to individuate others. Manipulation using digital models demonstrated that two species of cichlid fish use facial patterns, but not body colouration, to recognise familiar individuals (Kohda et al., 2015; Satoh et al., 2016). A species of reef fish uses UV patterns on the face for species recognition, but there is no evidence of IR (Siebeck et al., 2010). Medaka are shoaling fish with diverse social behaviours that has become a popular model in genetic and neural research. Medaka females prefer males with larger body sizes (Howard et al., 1998) and longer fins (Fujimoto et al., 2014), or familiar males. Visual contact for 5 hr can shorten the time to mate for a pair of medaka, and a certain extrahypothalamic neuromodulatory system alters the preference in response to familiarity (Okuyama et al., 2014). Nonetheless, the cues used for medaka IR and the cognitive basis that underlies IR remain unknown. Here, we investigated the identity signals used for medaka IR, and whether the process of recognising other individuals differs from that for other objects. We propose that medaka can become a powerful model for understanding IR systems for many reasons. First, abundant closely related species with different social behaviours are available, allowing us to test the evolutionary background that promotes strict IR. Second, the social behaviours within the species are also variable. Medaka from different geographic regions or different inbred strains behave uniquely (Tsuboko et al., 2014), allowing us to investigate how ecological factors influence the use of identity signals, as well as the mechanisms behind these signals. Moreover, rich genetic techniques such as genome editing and epigenetic methods are available for medaka (Kirchmaier et al., 2015), providing powerful tools with which to solve complex questions.

The first aim of this study was to identify the cues used for medaka IR. We tested whether visual and odour cues are part of the identity signals, and whether the cues work collaboratively. We also investigated which visual components (such as appearance, motion and different body parts) are necessary for IR, as well as the extent to which the signals can be manipulated (extra pattern added or image inverted) without affecting IR. The second aim was to test whether the mechanism of face recognition differs from that for non-face objects using the classic face-inversion paradigm and the accuracy of discriminating faces and non-face objects. We used both ecologically realistic settings (mating test) and a conditioned test (electric shock two-alternative forced-choice [TAFC] design) to assess strict IR in medaka. Understanding the cues that animals use to recognise others, as well as their cognitive basis, can help us to elucidate how animals connect to each other in their social world.