Understanding Fish Cognition

Animal Cognition (2021) 24:395–406 https://doi.org/10.1007/s10071-021-01488-2 REVIEW Understanding fsh cognition: a review and appraisal of current practices Matthew G. Salena1 · Andy J. Turko1,2,4 · Angad Singh3 · Avani Pathak1,2 · Emily Hughes2 · Culum Brown5 · Sigal Balshine1 Received: 15 June 2020 / Revised: 24 December 2020 / Accepted: 6 February 2021 / Published online: 17 February 2021 © The Author(s), under exclusive licence to Springer-Verlag GmbH, DE part of Springer Nature 2021 Abstract With over 30,000 recognized species, fshes exhibit an extraordinary variety of morphological, behavioural, and life-history traits. The feld of fsh cognition has grown markedly with numerous studies on fsh spatial navigation, numeracy, learning, decision-making, and even theory of mind. However, most cognitive research on fshes takes place in a highly controlled laboratory environment and it can therefore be difcult to determine whether fndings generalize to the ecology of wild fshes. Here, we summarize four prominent research areas in fsh cognition, highlighting some of the recent advances and key fndings. Next, we survey the literature, targeting these four areas, and quantify the nearly ubiquitous use of captive-bred individuals and a heavy reliance on lab-based research. We then discuss common practices that occur prior to experimenta- tion and within experiments that could hinder our ability to make more general conclusions about fsh cognition, and suggest possible solutions. By complementing ecologically relevant laboratory-based studies with in situ cognitive tests, we will gain further inroads toward unraveling how fshes learn and make decisions about food, mates, and territories. Keywords Teleosts · Memory · Intelligence · Learning · Behavior · Decision-making Introduction cognitive research. The more than 30,000 species of fshes provide valuable subjects for cognitive studies because of Over the last few decades, interest in the felds of animal their taxonomic diversity, variety of habitats, and range cognition and cognitive ecology has increased dramatically of life-history strategies (Patton and Braithwaite 2015). (Brown et al. 2011; Bshary and Brown 2014; Dukas 1998; Despite the growing interest in the cognitive ecology of Dukas and Ratclife 2009). Studies of animal cognition aim fshes, studies to date have focused on standard laboratory to understand the processes that help animals make deci- models, and only a handful of fsh cognition experiments sions (e.g., perception, learning, and memory; Ebbesson and have been conducted in the feld or on wild fsh. Captivity Braithwaite 2012). Fishes are well suited for such cognitive can have severe impacts on cognition both due to plasticity studies and have become regular experimental subjects in during an individual’s lifetime and via artifcial selection over multiple generations in the laboratory. Thus, the reli- * Matthew G. Salena ance on captive-bred fshes may limit our understanding of [email protected] fsh cognition in nature. Noting this strong laboratory bias and the accompanying gap in our knowledge, we embarked 1 Department of Psychology, Neuroscience and Behaviour, on this systematic survey and commentary of fsh cognition McMaster University, Hamilton, Ontario, Canada research. Our objective is to bring attention to the biases pre- 2 Department of Biology, McMaster University, Hamilton, sent in the literature and to encourage the thoughtful design Ontario, Canada of ecologically relevant experiments. Before discussing the 3 Department of Health Sciences, McMaster University, fndings of our literature survey, we frst provide a brief syn- Hamilton, Ontario, Canada opsis describing our current understanding of fsh cognition. 4 Great Lakes Institute for Environmental Research, University of Windsor, Windsor, Ontario, Canada 5 Department of Biological Sciences, Macquarie University, Sydney, Australia Vol.:(0123456789)1 3 396 Animal Cognition (2021) 24:395–406 What we know about fsh cognition process and has only seldom been demonstrated in fshes (Agrillo et al. 2009; Davis and Memmott 1982). The feld of fsh cognition has a reasonably long history, Many fshes use quantity assessment to inform eco- as comparative psychologists studied goldfsh (Carassius logically important behavioural decisions (e.g., what auratus) alongside rats and pigeons for over 100 years shoal to join, where to forage, or what mating tactic (Churchill Jr 1916). In the last decade, several excellent to use; reviewed by Agrillo et al. 2017). For exam- review papers have been published on the topic of fsh ple, fshes often choose to afliate with larger groups cognition (see Brown 2015; Bshary et al. 2014; Patton when given a choice between two diferent shoal sizes and Braithwaite 2015; Pouca and Brown 2018; Sneddon and there are numerous ftness benefts for doing so, and Brown 2020). The research has mainly focused on such as improved foraging and predator defense, four areas: (1) simple learning, (2) numeracy, (3) spatial increased vigilance, predator confusion, and dilution cognition, and (4) social cognition of fshes. For the pur- of risk (Agrillo et al. 2017). However, assessing the pose of this paper, we also focus on these four areas of fsh extent of more complex numerical abilities, such as cognition, summarizing current research in each area and counting, requires complex experimental protocols. then organizing our survey such that our literature search In mosquitofsh (Gambusia holbrooki), individuals targeted the practices in each area. showed a preference for larger shoals even when a series of bafes meant that only one conspecifc could i. Simple learning be viewed at any given time; thus, the focal ‘choosing’ Learning describes an animal’s ability to use infor- fsh needed to count how many individuals were at mation from past experiences to inform future behav- each end of the arena (Dadda et al. 2009). It seems that ior (Cauchoix and Chaine 2016). Simple learning true numerical representation by fshes is largely lim- includes non-associative forms of learning, such as ited to numbers no greater than 4 or 5, while ratios are habituation and sensitization. It also includes asso- typically used to compare larger quantities, consistent ciative forms of learning, in which connections are with many mammalian studies (Agrillo et al. 2017). made either between unconditioned and conditioned For instance, mosquitofsh discriminated between two stimuli (classical conditioning) or between stimuli and shoals that difered in number by a single individual a certain behavior (operant conditioning). when each shoal had fewer than 5 fsh, but discrimi- In fshes, simple learning can be rapid and long- nation between larger shoals was only possible if the lasting (Brown et al. 2011). For example, goldfsh bigger of the two had twice as many individuals or learned to avoid an area of a tank after a single electric more (Agrillo et al. 2008). shock (Riege and Cherkin 1971). Similarly, zebrafsh iii. Spatial cognition (Danio rerio) learned which colors predicted electric Spatial cognition is the ability to acquire and shocks with 89% accuracy after only 2 h of training reorganize spatial information to make sense of an (or 20 trials; Aoki et al. 2015). Crimson spotted rain- environment (Poucet 1993). Spatial cognition plays bowfsh (Melanotaenia duboulayi) greatly improved a role in many behavioural processes including for- their escape response to a novel trawl apparatus over 5 aging, mating, predator avoidance, and migration trials and they highlight the longevity that such asso- (Fukumori et al. 2010). Animals can navigate space ciations can be retained, remembering learned escape- using orientation (egocentric) or mapping (allocen- techniques for up to 11 months (Brown 2001). For tric) strategies and some fshes, such as goldfsh, rely many fshes, simple learning also begins early in life. on both (Rodriguez et al. 1994). Other species, like Zebrafsh, for example, can learn basic classical and the weakly electric elephantnose fsh Gnathonemus operant conditioning tasks from as early as 4 weeks of petersii, preferentially use egocentric cues during age (Valente et al. 2012). Simple learning in fshes is maze learning experiments (Schumacher et al. 2017). pertinent to survival-related tasks like predator avoid- In contrast, intertidal gobies (Bathygobius soporator) ance and foraging (Kiefer and Colgan 1992). create cognitive maps of the shoreline (thereby relying ii. Numerical cognition primarily on allocentric cues) and use these maps to Numerical cognition refers to the ability to dis- jump between nearby tide pools when threatened and criminate between two diferent discrete or continuous then can return to their home pool quickly (Aronson quantities (Agrillo et al. 2011). The ability to discern 1951; Jorge et al. 2012; White and Brown 2013). To quantities is widespread among vertebrates and some construct and use a cognitive map (i.e., a mental rep- invertebrates, while abstract numerical representation resentation of an environment), an animal needs to: (i) (counting) is considered a more demanding cognitive encode information about an object relative to other landmarks, (ii) integrate newly acquired information 1 3 Animal Cognition (2021) 24:395–406 397 into the map, and (iii) use the map to come up with carried out a PubMed subject search using seven search novel movement strategies (Poucet 1993). terms specifc to each of the four cognitive areas described iv. Social cognition above (a full list of search

Load more