Approach-avoidance conflict
The tension of wanting and dreading the same goal, which pulls you toward it from afar and pushes you back up close.
What it means
Approach-avoidance conflict is the motivational tension that arises when a single goal has both attractive and aversive qualities, so the same object elicits competing tendencies to approach and to withdraw. Lewin and Miller analyzed it with gradients: the avoidance gradient is steeper than the approach gradient, meaning the desire grows as the goal nears but the dread grows faster, so people advance until the two gradients cross and then stall or vacillate at that point. This explains chronic indecision, oscillation, and getting 'stuck' partway toward ambivalent goals like commitment, confrontation, or a tempting-but-risky opportunity. Related forms include approach-approach (two good options) and avoidance-avoidance (two bad options) conflicts. It matters as a precise mechanical account of ambivalence and why people freeze at a specific distance from goals they both want and fear.
Why avoidance climbs faster
Approach runs on an internal drive state, hunger or ambition or desire, that you carry with you rather than meet at the goal. Fear is a learned drive, conditioned to cues sitting at the goal itself. Those cues sharpen as you close in, so dread rises with them, while wanting rises too — but shallowly, because its source is a drive state you carry rather than a cue at the goal. The gradients differ in slope for a reason rooted in where each motive is stored: one in the person, one in the situation. So the crossing point is a property of the setting, not a defect of will, and it moves when the cues move.
What the evidence shows
Brown's rats, testing Miller's model, pulled against a harness in a runway; human work substitutes smaller movements. Boyd, Robinson and Fetterman had people move a joystick toward liked and away from disliked images with no correct direction, and across three experiments movement times were slower under avoidance conflict than approach conflict, consistent with, though not uniquely diagnostic of, the motor signature Miller predicted. Mouse-tracking sharpens this: Garcia-Guerrero and colleagues found avoidance responses first drift toward the reward before veering off, and trajectories grow more deflected, with more reversals, as the motives near equal strength. The vacillation is visible in the hand. Le Duc and colleagues report that people with hippocampal damage approach conflicted options more often, with a lower decision threshold and slower accumulation of evidence toward avoiding.
How it is measured
The animal lineage divides. The elevated plus maze and light-dark box trade exploratory drive against threat; the Vogel and Geller-Seifter punishment tasks go further and put a consummatory reward and a shock in the same place. Aupperle and colleagues reverse-translated this into a runway on a screen: an avatar moves between a positive image-and-sound worth no points, and a points-bearing option paired with an aversive image and sound (2, 4 or 6 points), and where it settles indexes the conflict. Reliability is uneven. McDermott and colleagues retested thirty healthy adults about two and a half weeks apart on average: approach behaviour held up well, with intraclass correlations near .77 to .80, while reaction time was only fair at .44 to .56, and the fMRI signal split by phase — the decision-making and reward phases were largely below fair (20.0% and 16.7% of ROIs at least fair), though affective outcomes held up better at 66.7%.
Limits and caveats
The gradients are inferred, never observed. In Miller's runway, distance was literal metres; in human life it is temporal or psychological, and treating those as one quantity is an extension the original data do not license. Heilizer's review made a sharper point that still stands: oscillation is conceptually the centre of the theory yet is rarely measured directly, so its most distinctive prediction is among its least tested. The steeper-avoidance asymmetry is itself an assumption fitted to particular reward and punishment pairings rather than a law. Treat the model as a shape that often fits ambivalence, not a constant to read values off.
Using it in practice
The model's most useful implication is counterintuitive: sweetening the reward is the weak move. A bigger incentive lifts the approach gradient fairly uniformly, shifting the crossing point only slightly nearer the goal and leaving the stall intact, so the person arrives closer and freezes anyway. Because the avoidance gradient is cue-bound and steep, removing aversive cues near the goal moves the crossing point much further per unit of effort. So decide at distance and precommit while wanting still dominates, shrink the aversive detail instead of inflating the prize, and read hesitation at the threshold as information about the last few metres, not about how much someone wants the thing.
Examples
Eager to propose marriage from a distance, a person grows anxious and hesitates as the moment actually approaches, hovering at the brink.
You want the raise and rehearse the ask all week, but outside your manager's door the dread outruns the wanting, so you hover, knock, and walk straight past.
The gym membership feels excellent on Sunday, when Monday is safely distant; at six the next morning, bag already packed by the door, the dread of the class wins.
A shopper fills a cart happily for twenty minutes, then stalls at the payment page where the fee, the card form and the commitment all sit together, and abandons within reach of buying.
A patient books an exposure session three weeks out with real resolve, then cancels an hour before it starts, once the feared cue is close enough for dread to outrun motivation.
First described in Kurt Lewin (1935); formalized by Neal Miller (1944).
Key references
- Le Duc, W., Butler, C. R., Argyropoulos, G. P. D., Chu, S., Hutcherson, C., Ruocco, A. C., Ito, R., & Lee, A. C. H. (2025). Hippocampal damage disrupts the latent decision-making processes underlying approach-avoidance conflict processing in humans. PLOS Biology, 23(2), e3003033. doi.org/10.1371/journal.pbio.3003033
- Garcia-Guerrero, S., O'Hora, D., Zgonnikov, A., & Scherbaum, S. (2022). The action dynamics of approach-avoidance conflict during decision-making. Quarterly Journal of Experimental Psychology, 76(1), 160-179. doi.org/10.1177/17470218221087625
- McDermott, T. J., Kirlic, N., Akeman, E., Touthang, J., Clausen, A. N., Kuplicki, R., & Aupperle, R. L. (2021). Test-retest reliability of approach-avoidance conflict decision-making during functional magnetic resonance imaging in healthy adults. Human Brain Mapping, 42(8), 2347-2361. doi.org/10.1002/hbm.25371
- Kirlic, N., Young, J., & Aupperle, R. L. (2017). Animal to human translational paradigms relevant for approach avoidance conflict decision making. Behaviour Research and Therapy, 96, 14-29. doi.org/10.1016/j.brat.2017.04.010
- Aupperle, R. L., Sullivan, S., Melrose, A. J., Paulus, M. P., & Stein, M. B. (2011). A reverse translational approach to quantify approach-avoidance conflict in humans. Behavioural Brain Research, 225(2), 455-463. doi.org/10.1016/j.bbr.2011.08.003
- Boyd, R. L., Robinson, M. D., & Fetterman, A. K. (2011). Miller (1944) revisited: Movement times in relation to approach and avoidance conflicts. Journal of Experimental Social Psychology, 47(6), 1192-1197. doi.org/10.1016/j.jesp.2011.04.017
- Miller, N. E. (1944). Experimental studies of conflict. In J. McV. Hunt (Ed.), Personality and the behavior disorders (Vol. 1, pp. 431-465). Ronald Press.
- Brown, J. S. (1948). Gradients of approach and avoidance responses and their relation to level of motivation. Journal of Comparative and Physiological Psychology, 41(6), 450-465.
- Heilizer, F. (1977). A review of theory and research on Miller's response competition (conflict) models. Journal of General Psychology, 97(2), 17-71.