When Rewards Stop, Does a Colour Keep Its Claim on Attention?

The coloured object can no longer earn you anything. It is not the target, and looking for it would not help you answer the question on the screen. Yet its earlier connection with money may still interfere with the answer you now need to give. A 2011 experiment by Brian Anderson, Patryk Laurent and Steven Yantis followed that change of roles: a useful target feature became an irrelevant distraction. The interesting question was not simply whether colour attracts attention. It was whether two similarly conspicuous colours would interfere differently because their reward histories had been different. [1]

Giving a feature a financial history

Eighteen people from the Johns Hopkins University community took part in the first experiment. During training, each display contained six differently coloured circles. The target was red or green, and participants reported whether the line inside it was horizontal or vertical. Finding the colour was necessary, but colour did not determine the response key. Either line orientation could appear in either target colour. This arrangement matters because an association between a colour and money need not be the same as a preference for making one particular rewarded hand movement.

Correct responses were followed by information about earnings. One target colour brought five cents on eighty per cent of trials and one cent on the remaining twenty per cent. The probabilities were reversed for the other colour. High value therefore meant a greater probability of the larger reward, not a guaranteed large payment every time that colour appeared. Red was the higher-value colour for half the participants; green held that role for the other half. The contrast was learned within the experiment rather than assigned to an assumed universal emotional meaning of red or green.

Training consisted of 1,008 trials. The colours in the later test consequently had a substantial history, rather than appearing for the first time when distraction was measured. During training itself, responses to the higher-reward target were slightly faster, but that difference was not statistically significant. It would be misleading to make a firmly demonstrated training-speed advantage the foundation of the story. The central result concerned what happened after the reward arrangement ended and participants were asked to search in a different way. [1]

Removing the reason to search for colour

In the test, shape defined the target. Participants found a white circle among white diamonds, or a white diamond among white circles, and again reported its internal line orientation. On half the trials, a nontarget object was red or green. The target itself was never either of those colours. Participants were told that colour was irrelevant and should be ignored. They no longer received monetary rewards, although they still received information about whether their answer was correct. The conspicuous colour now offered neither the right location nor a way to increase earnings.

There are two comparisons inside this arrangement, and they answer different questions. Comparing a coloured-distractor trial with a no-distractor trial measures interference from a colour that stands out against the otherwise white display. That difference alone cannot identify an effect of reward history. The more discriminating comparison is between the formerly high-value and formerly low-value distractors. Both had previously been targets. Both were physically distinctive at test. Their earlier probabilities of producing the larger reward supplied the crucial difference between their histories.

Both distractor colours slowed correct responses relative to trials without a coloured distractor. More importantly, the formerly high-value colour produced greater slowing than the formerly low-value colour. The planned statistical comparison supported that difference. Learned value thus added interference beyond the distraction shared by the conspicuous colours. This was a response-time result, not a record of where every participant's eyes moved. The experiment did not establish that each slower answer was accompanied by a visible glance at the distractor. Its direct evidence concerns the time needed to give the correct target response. [1]

A persistent effect need not be a permanent one

The test contained 480 trials, which the researchers examined in four successive groups of 120. The response-time difference between the high-value and low-value distractors declined as the unrewarded test continued. The earlier association did not disappear completely just because the instructions changed, but neither did it remain unchanged throughout subsequent experience. Both parts belong in the interpretation. Calling the result a lasting, unalterable preference would discard the very change that the repeated test trials made visible.

The decline also does not provide a fixed number of repetitions needed to remove a distraction in ordinary life. It is a group-level pattern within this task, not an individual recovery timetable. Error rates did not differ significantly between distractor conditions, but this is not proof that accuracy was exactly equivalent. The evidence was clearest for the timing pattern. Keeping response time and accuracy separate avoids turning one particular performance cost into a broader claim that participants became less capable of identifying the correct target altogether.

Could an old target explain the result without money?

The second experiment tested a plausible alternative. Perhaps an object attracted attention simply because it possessed a feature that participants had recently searched for. Eighteen new participants completed a similar procedure without monetary reward feedback. They received a fixed twenty-five dollars instead. Correctness feedback remained, so this was not a condition with no information about performance. During training, half searched for blue or red targets, while the other half searched for blue or green targets. This made one of the later red or green distractors a former target colour and the other a former nontarget colour. [1]

At test, both kinds of coloured distractor again slowed responses relative to having no coloured distractor. However, the researchers did not detect a response-time difference between the former-target and former-nontarget colours. This finding weighs against explaining the first experiment solely by a feature's earlier target status. It does not prove that target history can never matter in any task. The experiments also differed in their training arrangements and involved different people, so the control should not be described as a perfect isolation of every psychological ingredient associated with receiving a reward.

The authors additionally compared the high-value distraction in the first experiment with former-target distraction in the second. That cross-experiment difference supported the reward-history interpretation. It should not be mistaken for the high-minus-low reward contrast within the first group. These are different comparisons with different baselines. Keeping them distinct allows the control to strengthen the account without manufacturing a single effect size that describes every part of the design. None of these analyses required assuming that the participants consciously wanted to keep pursuing a colour that no longer helped them.

What happens during the extra time?

A delayed answer leaves more than one possible process open. Learned value might increase the priority given to an object before selection. Alternatively, once a conspicuous object has attracted attention, its earlier value might make disengaging from it slower. A previously rewarded search priority might also persist despite the new goal. The article discusses these possibilities rather than conclusively locating the effect at one stage. There were no eye-tracking or brain-imaging measurements here that could independently reveal the entire sequence hidden inside the response-time difference.

The resulting conclusion is specific but substantial. In these laboratory groups, a conspicuous colour with a history of larger probable rewards interfered more with a new, unrewarded shape task than a similarly conspicuous colour with a lower-value history. That additional interference diminished during further testing. The study does not establish a diagnostic test, a treatment, or a universal account of how notifications affect people outside the laboratory. Its contribution is the controlled change of roles: what once helped earn money later competed with the current task. Attention was influenced not only by the present display and present instruction, but also by the value that a feature had acquired along the way.

Sources

[1] Anderson, B. A., Laurent, P. A., and Yantis, S. (2011). Learned Value Magnifies Salience-Based Attentional Capture. PLOS ONE, 6(11), e27926. DOI: 10.1371/journal.pone.0027926. Creative Commons Attribution open-access primary research. Full main text reviewed on 15 October 2026. Independently written prose; no raw-data reanalysis performed.