Introduction

In animal memory research, a number of studies have been performed on animals to test for short-term or working memory (Vaughan and Greene 1984), whereas long-term memory studies are scarce due to the impracticality of testing requirements in the case of human subjects and the lack of effort exerted to understand long-term memory systems in non-human animals (Cook et al. 2005). Studies in non-human animals, such as birds, reported a large long-term memory capacity of pictures for more than 2 years in the most widely studied group of birds, White Carneaux pigeons (Vaughan and Greene 1984). In this study, the pigeons were not only able to recognize the 320 different pictures presented but also categorized pictures with no conceptually distinguishing characteristics into positive and negative categories. The detailed study of Cook et al. (2005), using a more cognitively demanding task (where the right or left response was arbitrarily assigned correct for a particular picture), supported Vaughan and Greene (1984) findings of the large long-term associative memory capacity of pigeons, by limiting the amount of information maintained at the same time.

In several families of the Passeriformes, the most recently evolved group of birds that include the corvids, cognitive abilities for successful retrieval of cached food that could be recovered hours, days, or months later are well documented (Emery and Clayton 2004). For example, Clark’s nut crackers can recover cached pine seeds after 9–11 months (Vander Wall and Hutchins 1983; Balda and Kamil 1992). Western scrub jays were able to recover cached food at intervals of 4–124 h (Clayton and Dickinson 1998), and magpies could recover them the same day or the next day after hoarding (Zinkivskay et al. 2009). Studies have also shown that magpies and western scrub jays can remember the ‘what’, ‘where’, and ‘when’ (episodic-like memory) of the food they hoarded after different time intervals (Zinkivskay et al. 2009). Recently, long-lasting memory of human faces for at least 2.7 years following a negative experience (crows were trapped and released wearing a particular face mask) is reported in wild American crows (Marzluf et al. 2010). These different studies indicate the survival value of retrieving biologically relevant information for later use.

In the case of jungle crows, an outstanding visual working memory capacity was reported using an operant matching-to-sample procedure (Goto and Watanabe 2009). In this study, jungle crows showed longer visual working memory capacity than pigeons; they were able to maintain above-chance performance at a delay of 64 s while the performance of the pigeons declined to chance-level at a 20 s delay (White 1985). However, data are still limited on the long-term memory capacity of jungle crows though spatial long-term memory of birds both in terms of time and amount of information stored at a time are abundant. Hence, using simple but ecologically relevant visual stimuli such as color, we tested how long jungle crows could retain and retrieve information using a learned color discrimination task in a semi-natural experimental setup. To the best of our knowledge, this study is the first to investigate long-term memory capacity of visual color stimuli in corvids. Remembering the rewarded color stimuli after long retention interval periods may indicate jungle crows’ larger long-term memory retention capacity and the biological significance of color cues.

Methods

Subjects and housing condition

Twenty jungle crows were caught by traps set in Niiza City and at the experimental farm of Utsunomiya University. The jungle crows were adults and determined to be at least 2 years of age based on the black color of the upper palate of the beak. Four crows each were assigned to 1-, 2-, 3-, 6-, and 10-month memory retention interval groups. Subjects were then marked with leg bands for individual identification and housed in groups of four individuals in an outdoor aviary (2.5 × 2.5 × 1.9 m). During the experiment, individual jungle crows were moved to an aviary of the same size where training and memory testing was conducted (Fig. 1). During each retention interval, crows were kept in groups of four in outdoor aviaries (2.4 × 2 × 2.5 m) that were placed in a different location from where the training was conducted. All of the aviaries contained natural vegetation, soil, stones, and a few perches and crows were allowed to fly freely.

Fig. 1
figure 1

Experimental cage and setup

Full size image

Subjects were provided food and water ad libitum throughout the course of the experiment. Their diet consisted mainly of full-nutrition dog food (Vita One, Nippon Pet Food Co., Tokyo, Japan). Water was freely available throughout the experimentation period. All of the jungle crows were wild-caught and thus were naive to the experiment of the two-alternative discrimination task. Participation of the crows was voluntary during the experimental trials. In the few cases where the crows did not want to participate, we stopped the session and resumed it the following day.

Stimuli

The stimuli were printouts (HP Deskjet D4160 printer, USA) of colors on a white paper (Full Color Paper 659-038, ASKUL, Japan). We used a single set of stimuli where two colors covered an equal portion in each stimulus (red with green and yellow with blue). The reflectance of the colors used was measured using spectrophotometer V-650, JASCO Corporation, Japan (Fig. 2) with deuterium and halogen lamps as light source. The color stimuli values were measured using Adobe Photoshop (Table 1). The printed stimuli were then pasted onto two cups, each 12 cm in diameter and 6 cm in height that were presented simultaneously.

Fig. 2
figure 2

Percentage reflectance of red (a), green (b), yellow (c), and blue (d) as a function of wavelength

Full size image
Table 1 Color stimuli values represented by primary hues
Full size table

Procedure

Pre-training

Crows were caught from the group aviary and kept individually in the experimental aviary (Fig. 1). The crow was provided with food and water and left for 1 or 2 days to become familiarized with the new aviary. The following day, the subject was presented with two white paper covered cups that contained dog food to habituate the bird to the experimental situation and to train the bird to peck the paper cover to obtain a food reward. Pre-training occurred for 1 or 2 consecutive days depending on the willingness of the subject after which pre-training was terminated. Training with the stimuli set commenced the following day.

Training

During training, two colored paper covered cups were simultaneously presented (Fig. 1). The correct choice was rewarded with two pieces of dog food while the incorrect choice led to the absence of reward and the experimenter enter to the cage and immediately removed the cups. The position of the stimuli-covered cups were changed randomly such that a single stimulus did not appear more than two times in the same position (right or left). The crows received ten trials in a session and one session per day. The discrimination training continued until the birds reached a criterion of significant discrimination, defined as a greater than or equal to 80% correct choice in two consecutive sessions, determined using a two-tailed binomial probability test (significance defined as p < 0.05). A video record was available throughout the experiment to record the response of the jungle crows. Training was conducted on clear-sky mornings, 10:00 to 11:00 a.m., covering autumn, winter, and spring seasons (Table 2). Once the jungle crows completed the training, they were kept in separate outdoor aviaries for 1-, 2-, 3-, 6-, and 10-month retention intervals, until they were tested for memory.

Table 2 Training and memory test periods of individual crows
Full size table

Memory test

The jungle crows were tested for retention of discrimination performance after delays of 1, 2, 3, 6, and 10 months without seeing the training stimuli during the prescribed time. The crows were also kept in different aviaries in a different location than that used during training. During this period, the crows were not involved in any other experiments. The memory test was conducted in the same aviary where the crows were trained. Significant discrimination criterion of nine correct selections out of ten trials (90% correct choice) in a single session was set (two-tailed binomial probability test, significance defined as p < 0.05). The testing procedure was similar to the training phase except that the crows received only one session of ten trials. Testing was conducted on clear-sky mornings between 10:00 and 11:00 a.m. as in training phases during winter, spring, and autumn seasons (Table 2).

Data analysis

The two-tailed binomial probability test was used to determine discriminative behavior between the two color stimuli in jungle crows.

Results

The total number of trials that the crows required to reach discrimination criterion ranged from 20 to 40 trials (mean: 32.5 ± 6.15). Some crows reached the discrimination criterion within a single session (crows 2 and 6) while most of the crows reached the criterion in the second session (12 crows) (Fig. 3).

Fig. 3
figure 3

Discrimination performance of jungle crows during training shown for each 1 (a), 2 (b), 3 (c), 6 (d) and 10 (e) month retention interval group. The broken line indicates the minimum significant performance to reach criterion (two-tailed binomial probability test, p < 0.05)

Full size image

Individual crows’ response in the memory test showed that most jungle crows significantly discriminated the previously constantly reinforced stimuli and made no or few mistakes (two-tailed binomial probability test, p < 0.05). One bird failed the memory test in each of the 1 (crow 2, p = 0.054), 3 (crow 9, p = 0.62) and 10 (crow 2, p = 0.17) month retention intervals while two crows (crows 15 and 16) failed to remember the reinforced stimuli during training after a 6-month retention interval (p = 0.37 in both subjects), two-tailed binomial probability test (Fig. 4). The performance of all crows (except crow 18 that performed at 90%) that discriminated the correct reinforced stimuli during memory test was 100%. In addition, selection of the correct stimuli just in the first trial in the 20 crows was significant (two-tailed binomial probability test, p = 0.0002) during the memory test.

Fig. 4
figure 4

Discrimination performance of jungle crows during the memory test after 1-, 2-, 3-, 6-, and 10-month retention intervals. The broken line indicates the minimum significant performance (two-tailed binomial probability test, p < 0.05)

Full size image

Discussion

We investigated the ability of jungle crows to discriminate a reinforced visual stimuli after a given retention period without having been exposed to the stimuli in between. The jungle crows were able to discriminate between the color stimuli and reached a discrimination criterion using only a few trials. Some crows were able to reach the criterion within a single session, indicating that crows can easily learn color discrimination tasks. A similar observation was also reported in ravens (Range et al. 2008).

The memory test performed after different retention periods showed a remarkable ability of jungle crows to retrieve learned information. The discrimination performance of the majority of the crows was not disrupted even after 10 months without seeing the stimuli. The performance of jungle crows did not decline with increasing retention intervals, indicating that their memory is resistant to decay, at least during this period. The decline in performance in some of the subjects after 1, 3, 6, and 10 months of retention intervals might be associated with individual differences in memory retention capacity that led to decay rather than an overall decay of memory over time (one through 10 months). In fact, jungle crows showed a decay of visual memory in the delayed matching-to-sample task in which retrieval was affected by intertrial proactive interference (Goto and Watanabe 2009), whereas visual long-term memory following training seems to remain preserved for a long period. One can reasonably argue that this capacity is retained due to the fact that the crows did not engage in other behavioral activity during the retention time. On the other hand, they also did not have an opportunity to embed the color rule in any context or reinforce it with other information such as the tastiness of one stimulus over the other during this keeping period. Despite these conditions, the crows remembered the rewarded stimulus, showing their long-term memory retention capacity.

Consequently, the outstanding performance of jungle crows may be due to the fact that color seems to be an important component in their visual environment for exploitation of biological resources such as food. Previous reports have shown that jungle crows’ diet is mainly constituted by fruits (Kurosawa et al. 2003), and the perception of chromatic (wavelength-related) contrasts in frugivorous birds was reported to be significant for fruit detection (Cazetta et al. 2009). Successful detection of a cryptic prey was also reported to depend on color contrasts in blue jays (Cyanocitta cristata) (Pietrewicz and Kamil 1977; Kono et al. 1998). Recently, color cues were reported as a major feature used to categorize human faces according to sex in jungle crows (Bogale et al. 2011), where this ability is significant for profitable exploitation of anthropogenic resources while co-existing in urban areas as was observed in American crows (Marzluf et al. 2010).

Mettke-Hofmann and Gwinner (2003) stated that an animal’s exposure to specific ecological conditions could shape its cognitive abilities. For example, long-lasting memory of a particular feeding site may increase survival in long-distance migratory birds such as garden warblers (Sylvia borin), saving time and energy during subsequent journeys. Any animal that lives in a highly seasonal environment, even if it does not migrate, memory capacity to recall seasonal food sources and to recheck them at appropriate times might be likely. For example, primates such as free-ranging Japanese macaques (Macaca fuscata) searched for artificially introduced seasonal fruits based on past memory rather than recent location information of finds (Menzel 1991). In food-caching corvids, recovery of the cached food appears to be directed by visual cues in the environment (Vander Wall 1982). Although their dependency may not be high, jungle crows have been reported to cache food items (Higuchi and Morishita 1997). Hence, such long-term memory of color stimuli might be associated with the ability to successfully retrieve cached food that might require memory of both physical and social contexts during caching (Emery and Clayton 2004). Caching provides one context for specific location memory and construction of more complex association between color, place, and food that may be essential in remembering food sources when territory size and time separate an animal from its food source for a long time, as was observed in Japanese macaques (Menzel 1991).

Previous studies showed that pigeons can be trained with as many as 160 pairs of photographs and random shapes (Vaughan and Greene 1984) and mammals such as sheep (Ovis aries) can recognize 50 familiar conspecific faces for over 2 years (Kendrick et al. 2001). Although we studied memory duration of simple visual information using only one pair of exemplar, the outstanding performance of jungle crows suggests that the long-term memory retention capacity of visual stimuli might be shared across species. The consistent memory capacity to relatively shorter and longer retention periods supported by the significant selection of the correct stimuli just in the first trials during tests imply memory rather than rehearsal, often observed in long-term memory studies of non-human animals. Further study is required regarding the amount of information that could be stored at a time in jungle crows.