A “mystery” oriole and the limitations of identification by impression

On 9 May 2011 several birders saw and heard a strange oriole in Georgetown, South Carolina (photos here). They identified it as a Scott’s Oriole, a first state record and the first spring record in the east. When other birders saw the photos, some questioned why the bird was not an Orchard Oriole (common in South Carolina) and the debate reached the ID-Frontiers listserve, where it generated mixed opinions and became one of the most evenly polarized identification questions ever considered there.

I am now confident that the bird is an Orchard Oriole based on plumage, and this sound recording by Ritch Lilly helps to close the case: Audio recording of the singing oriole (Thanks to Ritch for having the foresight to make the recording with his phone; and for sharing it).

Even though the identification now seems to be resolved, I think it’s worth looking back at this episode and examining some of the identification challenges it raises.

Identification: Orchard vs. Scott’s Oriole

Plumage offers the best visual clues to why this is an Orchard and not a Scott’s Oriole. Key points in favor of Orchard include:

  • The clean and sharp border between black throat/face and greenish cheeks and crown (Scott’s should have a more mottled border, with lots of black on the cheeks and crown, and it’s possible that Scott’s never shows a clean and contrasting “black-faced” pattern like this bird)
  • Greenish back without dark mottling (Scott’s should have a grayish back with obvious dark feather centers; the second photo of the SC bird shows this, other photos are ambiguous)
  • Narrow white wingbars, especially the lower (greater coverts) bar (Scott’s should show broader and more obvious white bars, both about equal in width – but this needs confirmation, it’s possible that some immature Scott’s have weak wingbars)

With that said, there are a number of things in the photos that say “Scott’s Oriole” at first glance. The dusky-olive head contrasting sharply with the yellow belly, and the appearance of a dark and mottled back (the mottling probably due to shadows) both create a strong impression of Scott’s Oriole that could easily be misleading.

Other features such as intensity of yellow underparts, extent of dark hood, tail pattern, etc. are suggestive for one species or the other but are simply not shown clearly enough in the photos. I don’t have much confidence in distinguishing these species by shape, partly because details are hard to make out in these photos, and partly because I just don’t know if the apparent differences are reliable.

Psychology

“Perception may be regarded as primarily the modification of an anticipation.”
Ernst Gombrich, Art and Illusion. 1956

The online responses to identification of the South Carolina oriole are notable for their emphasis on shape, and for their contradictory results:

“the structure of this bird is clearly Scott’s Oriole. Big body, big head, straight culmen, big tail.”

“The bird feels rather delicate. In particular, the head/bill size and shape look like Orchard to me.”

“the shape (jizz) of the SC bird struck me as “un-Orchard-like.”

“this bird looks fine for Orchard Oriole.  The bill looks too short and the body too slender for Scott’s.”

“it does not resemble Orchard Oriole’s physical structure, with a bigger head and longer body shape”

One problem with using shape here is that – unlike hawks, shorebirds, and other species for which shape is a primary field mark that has been refined over decades of study – few birders have ever compared these two orioles side-by-side, and the shape differences and variation are not well-understood. A second, more interesting, problem is that our visual assessment of objects is strongly influenced by expectations, and we will subconsciously tamper with the evidence to confirm a hypothesis. Thinking “It looks like a Scott’s” creates a bias and influences all later evaluation.

This skewing of our perception happens partly because of a tendency to overvalue evidence that supports our hypothesis and to undervalue evidence that refutes it. It also happens because our brains literally make adjustments to the data to match our expectations. That is, if we think the bird is a Scott’s Oriole, we will tend to perceive it as more Scott’s-Oriole-like.

A study by Duncker (1938) tested perception of color using familiar objects. A leaf silhouette was consistently perceived as more green than a donkey silhouette, even though the two were actually the same color. Oikkonen et al (2008) found a similar result in more sophisticated tests and wrote “the visual identity of an object has a measurable effect on color perception”. Our brains analyze the information coming from our eyes, and “fudge the data” so that the results more closely match past experience (accepting the green leaf and rejecting the green donkey).

Some of the colors and shapes that we “see” have been altered by our brains to match the expectations we’ve set up. We “see” what we think we should be seeing.

In evaluating colors on the South Carolina photos, observers who started with a hypothesis of Scott’s would not only accept the yellow belly as a supporting clue, but would also perceive the yellow as a brighter color. In the same way that we are predisposed to perceive a leaf as green, we know a Scott’s Oriole is supposed to be yellow, and therefore it looks yellow.

I suspect that in the case of these oriole photos, most viewers had an instantaneous subjective “feeling” that it was either Orchard or Scott’s, based on a gestalt impression of shape, posture, plumage, and intangible factors. With a strong sense of the species involved, studying the bird then leads each observer to perceive the shape of “their” species. Those who started with the hypothesis of Scott’s saw a big-headed, large-billed, large-tailed bird, and those who started with Orchard saw the opposite. Thinking it was a Scott’s Oriole would not only cause one to latch onto the details of shape that supported that conclusion, but would actually cause the brain to perceive the shape as more Scott’s-like.

In fact, the fundamental thing leading each observer to their conclusion could be some undefinable combination of subtle impressions such as color patterns on head and body; just enough to trigger the initial sense that this is a Scott’s/Orchard Oriole. That sense then triggers a series of other expectations and guides the study of the bird, leading to an identification that is ostensibly based on a careful assessment of shape, but in reality still rests mainly on the very weak foundation of vague first impressions.

To me, the first (top) photo on the Carolinabirds web page is the most Scott’s-like of the three images there. The last photo in the series appears to me the most Orchard-like. If the order of the photos had been reversed, would everyone’s first impression have been different, and would the debate have proceeded differently?

These are subtle effects, but they are operating all the time. Expectations allow us to identify things quickly and with a high degree of accuracy, but expectations can also trap us. Most bird sightings are not very ambiguous, but we need to recognize when a sighting is problematic, keep an open mind, and be especially vigilant for psychological effects that might lead us astray or give us false confidence.

Addendum (20 May 2011)

An interesting question from John Idzikowski: he wondered if there was a regional bias in this oriole debate. I checked all the posted opinions on ID-Frontiers and a couple from Carolinabirds, looking for a link between the species the observer was familiar with and the species they chose for the SC oriole.

Ten people were more familiar with Orchard Oriole, and those were evenly split, 5 for each species. Only three people reported being more familiar with Scott’s Oriole, and all three voted for Scott’s.

So… there was no clear regional or familiarity bias in the results. It’s a small sample and each one was influenced by all kinds of things, so there’s not much more that can be said about this. One possible explanation for the lack of a pattern is that people based their opinions on a wide range of evidence. Some focused on shape, some on plumage; some weighed the rarity of the bird more heavily, some weighed the testimony of the original observers more heavily, etc.

References

Duncker, K. 1939. The Influence of Past Experience upon Perceptual Properties. The American Journal of Psychology. 52: 255-265.

Olkkonen, M., Hansen, T., & Gegenfurtner, K. R. (2008). Color appearance of familiar objects: Effects of object shape, texture, and illumination changes. Journal of Vision, 8(5):13, 1–16.  http://www.journalofvision.org/content/8/5/13.short

8 thoughts on “A “mystery” oriole and the limitations of identification by impression”

  1. I agree with your analysis and conclusions but I think we lose sight of how limited their application might be. Most of what you say applies to static images, still photos certainly but as well perched or otherwise immobile birds. In the field, and in most situations, the amount of information that flows through to us is vastly increased and not only includes a wealth of behavioral cues but a seamless flow of images which allow us to refine and on occasion alter our original impressions.

    When looking at bird images, I have often found myself firmly and unhappily in the grasp of an “Idée fixe” but that rarely happens in the field even though the final identification is sometimes different from what it was in the first instant.

    1. Hi Will. Good point. I agree that these kinds of problems are much more common when viewing still photos, and especially lower-quality photos such as the SC oriole, but it can still happen in the field and the mechanisms are the same whether we’re looking at photos or live birds. If we see a bird well we do get a wealth of information and behavioral clues that photos do not provide. The real obstacle is limited information (which leads to ambiguity). Information can be limited by having only a few low-quality photos, or by getting a very brief or distant view of a live bird, or even in a good view by biases that cause us to take in only “pre-approved” information.
      I think these effects are common, and just unnoticed. I would propose that experienced birders subconsciously learn how to avoid the pitfalls and when to hold back on an ID, even if they don’t know what the pitfalls are.

  2. I loved every minute of this conversation and think that all of the points about the ID of this bird where valid, interesting and a very good learning experience. I think you came full circle on the whole subject of why and what we should always do, break it down and take a closer look at the details. The details along with a few other pictures made it seem easy in the end and I’d like to thank all who cared to but forth an opinion on the subject. It’s what it’s all about!

  3. I agree with Will’s comments and David’s reply, but would like to add a couple of points. First, I did not weigh in on the bird’s identity because I have only seen a few Scott’s Orioles, and its been quite a while since I have seen Orchard Orioles.

    I use size A LOT in identification of live birds. I have trained myself to be pretty good at recognizing size differences among birds: I generally know whether a female finch is Purple or House by size and shape before I look at plumage and bill characters. The most common Red Crossbills around my home on the Oregon Coast are the small “Sitka” form, and when visitors of different types show up, I first notice the larger size… etc. Further, my assessment of size is critical to my recognition and appreciation of structural differences.

    So, in photos I often have difficulty interpreting structural attributes if I cannot accurately estimate size. In a lot of bird groups, different sized relatives differ in proportions in fairly consistent ways, and when a species breaks these allometric patterns it can be really conspicuous. Thus, among Buteos, head size variation is fairly unremarkable, except Ferruginous Hawks have very wide heads that stand out like sore thumbs.

    Among Catharus thrushes, size and related proportions also vary in really useful (for i.d.) ways.

    I think we tend to overestimate the difficulties associated with i.d. from photos when objective plumage color features are not decisive.

  4. I also agree with Will’s points about the pitfalls of identifying birds from photographs and was surprised this really didn’t come up in the ID-Frontiers discussion. My guess is all of the folks who offered opinions on ID-Frontiers on the photos would have got the id correct in the field. And like Wayne I also had trouble judging the size of this bird from the photograph – it seemed very ambiguous to me. I agree that these problems can happen in the field but with care can be avoided (especially in the case of rarities). This case should also give pause to BRCs which more and more rely largely on photos to make their decisions.

  5. I would add the “null hypothesis” type of thinking to your discussion. Let’s say a VERY similar first-spring male oriole was singing from a Joshua tree in eastern California, and several folks identified it as an Orchard, even claiming they heard it sing. Now, IF Scott’s were a common breeder in that area (I don’t know if that is so), then one should start with the assumption that the bird is a Scott’s, and it is thus a requisite for the observers to dis-prove Scott’s, as much as prove it is Orchard.

    That was my basis all along on the SC bird, as I have NEVER seen a first-spring Scott’s. “Why isn’t this just an Orchard?” I kept asking. And, I stated that there must have been a 99.9% chance it was an Orchard over Scott’s, based on range, singing as if on territory, from a wetland area, etc. These factors didn’t scream Scott’s to me. Thankfully, I was right, after you first heard the song, and I heard it later.

    Thanks for your comments on the plumage, saying Orchard based on that alone, and not basing your comments on geography, habitat, etc., as I had to do, as I’m no expert on these plumages.

  6. I agree with the conclusion reached on this matter, but also think David Muth’s point made over at ID Frontiers is worth repeating here, that from a strictly technical viewpoint the answer isn’t 100% nailed down since we don’t know that the bird on tape and the photo’d bird are in fact one-and-the-same (for that we need a live video), but simply a strong assumption; small point, but worth acknowledging.

  7. The scotts are beautiful but are killing my hummingbird colonies. How can I keep them away from my hummingbird feeders?

Leave a Reply to Will Russell Cancel Reply

Your email address will not be published. Required fields are marked *