A Method for Investigating Change Blindness in Pigeons (Columba Livia)

Summary

Change blindness is a phenomenon of visual attention, whereby changes to a visual display go unnoticed under certain specific circumstances. This protocol describes a variation on the flicker paradigm for investigating change blindness that is appropriate and effective for research with pigeons.

Abstract

Change blindness is a phenomenon of visual attention, whereby changes to a visual display go unnoticed under certain specific circumstances. While many laboratory procedures have been developed that produce change blindness in humans, the flicker paradigm has emerged as a particularly effective method. In the flicker paradigm, two visual displays are presented in alternation with one another. If successive displays are separated by a short inter-stimulus interval (ISI), change detection is impaired. The simplicity of the procedure and the clear, performance-based operational definition of change blindness make the flicker paradigm well-suited to comparative research using nonhuman animals. Indeed, a variant has been developed that can be implemented in operant chambers to study change blindness in pigeons. Results indicate that pigeons, like humans, are worse at detecting the location of a change if two consecutive displays are separated in time by a blank ISI. Furthermore, pigeons' change detection is consistent with an active, location-by-location search process that requires selective attention. The flicker task thus has the potential to contribute to investigations of the dynamics of pigeons' selective spatial attention in comparison to humans. It also illustrates that the phenomenon of change blindness is not exclusive to humans' visual perception, but may instead be a general consequence of selective attention. Finally, while the useful aspects of attention are widely appreciated and understood, it is also important to acknowledge that they may be accompanied by specific imperfections such as change blindness, and that these imperfections have consequences across a wide range of contexts.

Introduction

Cognitive psychology has repeatedly demonstrated striking and often surprising imperfections in our own cognitive processes. Some of the more notable examples include but are not limited to false memories¹, suboptimal decision heuristics², and faulty statistical reasoning³. A more recent addition to this list is the phenomenon of change blindness. Change blindness is a consistent failure of attention, in which one fails to notice even prominent changes to one's environment. In one demonstration⁴, experimenters had a confederate approach individuals to request directions. During their conversation, workers carrying a door passed between the two, briefly interrupting visual contact and providing an opportunity to swap out the confederate for a different person. After this surreptitious exchange, most individuals failed to notice that their conversation partner was no longer the same person. This failure is surprising because moment-to-moment changes would seem to be signals of potentially important events that ought to draw our attention.

In order to better understand how and why change blindness occurs, researchers have brought it into the lab and developed several ingenious procedures^5,6,7,8 for studying it under more controlled conditions. One particularly successful approach has been dubbed "the flicker task"⁹. In this procedure, the experimenters edited photographs by removing a feature, and then presented the images to participants, rapidly alternating between the original and modified versions. Participants quickly spotted the differences. However, if a brief blank field was inserted between consecutive images (producing a flickering display for which the procedure is named) change detection was much more difficult, resulting in worse accuracy and slower response times. This procedure is appealing because it provides a precise measure of change blindness, and it is easy to manipulate specific aspects of the display such as the size, salience, or timing of a change.

The flicker paradigm has proven to be a powerful tool for learning about perception and attention in humans¹⁰. The effect is surprisingly powerful and persistent. Change blindness can occur for changes to the only object in a simple animation¹¹, and when looking directly at a change location¹². Even experience with change blindness and awareness of the phenomenon does not eliminate it¹³. Furthermore, change blindness is quite diverse, and can be induced by a number of different events, such as eye saccades⁵, mudsplashes¹⁴, motion picture cuts⁷, or visual occlusion⁴. A parallel phenomenon occurs in auditory¹⁵ and tactile¹⁶ modalities, indicating that it may not be exclusive to visual stimuli and may be a more general phenomenon of attention.

Humans of course, are not the only animal that makes use of attention. Pigeons, for example, show many of the same attentional abilities that humans do. They can select specific features for preferential processing (as when they use a search image to find specific food targets)^17,18. They can direct attention toward specific regions or spatial locations¹⁹. They can shift their attention between hierarchical levels of organization^20,21. They can also divide attention between different aspects of a stimulus display^22,23. It seems then, that pigeons possess many of the same abilities that make attention useful for humans. If change blindness has to do with some of the inherent limitations of attention, we might expect to see a parallel change blindness effect in pigeons (and perhaps other animals). Furthermore, there have recently been multiple successful studies of change detection conducted using pigeons, implementing widely varying methods^{24,25,26,27,28}.

Recent research^29,30,31,32 has adapted the flicker paradigm to investigate change blindness in pigeons, and has produced results that parallel human change blindness. The current report describes a simple method for implementing this procedure in an operant chamber. As with human versions of the task, it is easy to isolate and manipulate specific aspects of stimulus presentation in order to investigate their influence over change detection and change blindness. Thus, the procedure should constitute a valuable tool for research on the limitations of avian attention, and the extent to which they are similar to the limitations of human attention.

Protocol

The procedure described here is in accordance with the Office of Laboratory Animal Welfare and with US Public Health Service Policy on Humane Care and Use of Laboratory Animals, and was approved by Whitman College's Institutional Animal Care and Use Committee.

1. Reduce Pigeons' Weights

NOTE: Pigeons' weights are reduced to 80 - 85% of their free-feeding weight³³ to ensure that the birds are healthy and adequately motivated to work for food.

1. House naïve birds in individual cages with unlimited access to water, grit, and food.
2. Weigh each pigeon at approximately the same time each day for 2 to 4 weeks, or until each bird's free-feeding weight has stabilized.
3. Calculate a target weight for each pigeon equal to 85% of its stable free-feeding weight.
4. Restrict food to gradually reduce each pigeon's weight until the target weight is reached³⁴. Pigeons should still have unrestricted access to water and grit.
   NOTE: Published experiments using this protocol^29,30,31,32 have yielded significant results using between 4 and 6 pigeons per condition. Similar numbers should be adequate for a direct replication or subtle variation. Variations that reduce the magnitude of the change blindness effect could require a larger sample.

2. Train Pigeons to Peck Stimuli Displayed on the Response Keys in the Operant Chamber and to Eat Grain from the Food Hopper

NOTE: Training and experimental sessions require precise computer control, with temporal resolution of less than 1 ms. Use a flexible programming language (see Table of Materials) to control operant chambers via an I/O relay.

1. At the beginning of each day's session, weigh pigeons and place them into operant chambers (see Table of Materials). Naïve pigeons may need time to acclimate to handling, weighing, and transport to and from the operant chamber.  Until then, take extra care to handle birds gently and monitor them for signs of stress.
2. Run 100 trials per daily session. For each trial:
   1. Randomly select a visual stimulus element (such as a color or line) and one of the three keys (see Table of Materials) in the operant chamber. Illuminate the selected stimulus element on the relevant key using a compatible stimulus projector (see Table of Materials).
      NOTE: The onset and offset times of the incandescent bulbs that are standard in many operant chambers are too slow to be appropriate for this method. Replace any incandescent bulbs in the operant chamber with a faster LED equivalent, and then confirm that the displays appear as intended and that the onset of stimuli is crisp (less than 1 ms from onset to peak brightness).
   2. Wait until a pigeon pecks the key on which the stimulus is displayed. Do not acknowledge pecks to any other keys.
      NOTE: Experienced pigeons may immediately know to peck illuminated response keys. Naïve or less experienced birds' pecking can be shaped using handshaping or autoshaping³⁵ procedures as they would be for other laboratory tasks.
   3. Following a single peck to the proper response key, clear the stimulus display and provide access to grain from the food hopper for 2 - 3 seconds.
3. At the end of each session, remove pigeons from operant chambers and weigh them before returning them to their home cages. Adjust food access time between sessions to maintain birds' individual running weights at 80 - 85% of their free-feeding weights.
4. Continue pre-training sessions until pigeons respond quickly and consistently to all of the individual stimulus elements to be included in the experiment, on all three response keys.

3. Train Pigeons to Search for and Peck Changes Presented on Response Keys

1. At the beginning of each day's session, weigh pigeons and place them into operant chambers.
2. At the beginning of each of 100 trials in a daily session, determine the details of the upcoming stimulus display. Details for each trial can be randomly selected by the experimental control software. A sample program to run a daily experimental session is included as a supplemental file (change.cpp); elements present there could also be used to perform the simpler actions in step 2.
   1. Randomly choose an Inter-Stimulus Interval (ISI) of either 250 ms or 0 ms (with p = 0.5 for each).
   2. Randomly determine the number of change repetitions to present, either 1, 2, 4, 8, or 16 (with p = 0.2 for each).
   3. Define an original stimulus display consisting of one or more elements (the colors or lines presented during pretraining) on each response key.
   4. Change the original stimulus display to define a modified display by adding, deleting, or changing one element on one key. See Figure 1 for examples of original and modified stimulus displays.
   5. Ensure that pigeons will not see all possible stimulus displays during training. If necessary, designate a subset of displays for transfer (step 4) and refrain from presenting them during training.
3. Present a 5 s Inter-Trial Interval (ITI), with the houselight on and all response keys dark to separate each trial from the immediately preceding trial.
4. Present a stimulus display using the values determined for the current trial in step 3.2.
   1. Illuminate the stimulus elements that compose the original display for 250 ms.
   2. Clear the stimulus display and wait for an ISI of either 0 or 250 ms.
   3. Illuminate the stimulus elements that compose the modified display for 250 ms.
   4. Clear the stimulus display and wait for an ISI of either 0 or 250 ms.
   5. Repeat steps 3.4.1 to 3.4.4 until completion of the number of repetitions previously determined for the current trial. Present all repetitions in their entirety and ignore any keypecks during stimulus presentation.
5. Illuminate all three response keys with white light and wait until a pigeon pecks one of the three response keys. Consider the first peck on any response key after stimulus presentation is complete to be the response for that trial. See Figure 2 for a schematic description of trials with or without an ISI.
6. Clear all keys and conclude the trial with either reinforcement or an error signal:
   1. If a bird's response was on the key that displayed a change, provide access to grain from the food hopper for 2 - 3 s.
   2. If a bird's response was not on the key that changed, switch the houselight between on and off every 0.5 s for 10 seconds to indicate an incorrect response.
7. At the end of each session, remove pigeons from operant chambers and weigh them before returning them to their home cages. Adjust food access time between sessions to maintain birds' individual running weights at 80 - 85% of their free-feeding weights.
8. Continue daily training sessions until the accuracy of pigeons' responses is stable, and reliably better than chance accuracy of 33%. A sample file is provided (change.xlsx) that analyzes raw data to show the effects of ISI presence and number of repetitions.

4. Present Novel Transfer Trials Within Daily Sessions

1. Follow the procedure exactly as outlined in step 3, but without any potential displays excluded (see step 3.2.5).
   NOTE: With large numbers of potential stimulus displays, it is not necessary to exclude potential displays during training and introduce them later. In those cases, simply continue training as normal, as never-before-seen displays will occur naturally.
2. Analyze accuracy on novel, never-before-seen stimulus displays, excluding any displays pigeons have previously encountered. Better than chance accuracy will confirm that pigeons have learned a general change detection rule and are not relying on memorization of familiar stimulus displays.

Results

The primary result of interest is the difference in accuracy between trials with and without an ISI. In particular, the operational definition of change blindness in the flicker paradigm is significantly reduced change-detection accuracy on trials with an ISI relative to trials without an ISI. This effect can be seen in Figure 3, which shows previously published data²⁹. In that experiment, pigeons detected changes to stimuli consisting...

Discussion

The method presented here is inspired by the so-called "flicker paradigm" commonly used by cognitive psychologists to study change blindness in humans⁹. In this human research, change blindness is operationally defined as the impairment in change detection produced by the presence of an ISI that interrupts transitions between stimulus displays. The same is true for the pigeon implementation described here. Furthermore, humans tend to approach the flicker task using a serial search strategy...

Materials

Name	Company	Catalog Number	Comments
Small Environment Cublicle	BRS/LVE	SEC-002
Pigeon Intelligence Panel	BRS/LVE	PIP-016
Grain Feeder	BRS/LVE	GFM-001
Pigeon Pecking Key	BRS/LVE	PPK-001
Stimulus projector	BRS/LVE	IC-901
LED Lamp	Martek Industries, Cherry Hill NJ	1820
I/O module	Acces IO	USB-IDIO-8
Personal Computer	Dell	Optiplex 3040
Visual C++	Microsoft
White Carneau pigeons	Double-T Farm