Making it easier for players to form emotional connections with NPCs, both positive and negative, helps player retention. This article explores how we can use psychology to strengthen players' emotional bonds with NPCs in a singleplayer, team-based setting, including practical tips at the end. Through using intergroup empathy bias in NPC design we could give players a more fulfilling experience and keep them playing for longer.
How many games can you think of that portray conflict between two equal teams with individual, recurring characters? The genre of multiplayer games is full of these types of experiences, but to find a game like this in a singleplayer setting should be much harder. Sure, games that portray conflict between groups of Non-Player-Characters exist, yet examples like XCOM and RimWorld both feature opposing forces that vastly outnumber the player's team. A game where the recurring NPCs of the opposing team have similar strength should be hard to find, because the benefit is not directly obvious. Nevertheless, introducing this type of intergroup conflict into a game could make it easier to reach certain design goals:
Creating even stronger bonds between players and NPCs, and in turn, enhancing player-enjoyment and retention.
A core part of forming a parasocial relationship (i.e. a perceived relationship with a media character) is the creation of emotions in the player. This is both true for handcrafted characters in linear story-games like The Witcher 3, as well as in semi-procedural squad-based games like RimWorld and XCOM. Forming such a relationship can manifest itself as empathy and even counter-empathy (i.e. feeling good about someone else's misfortune). Both feelings can lead to different kinds of connections, but also to a stronger fulfillment of a player's need for relatedness.
This, according to the PENS-model (Player Experience of Need Satisfaction), should increase player-enjoyment and -retention. Scott Rigby, a game researcher, and his colleagues developed it to study what player-needs should be fulfilled in order to make games more enjoyable. If that model is accurate, then fulfilling the player's need for relatedness through NPC bonding will improve their experience within the game and keep them around for longer.
XCOM 2 uses customizable, randomized, mortal NPCs for the player to bond with
Group empathy bias, where players may like or dislike other characters based on their group affiliation in a conflict situation, has been widely studied in non-game contexts. For example, people tend to show more empathy towards their own group compared to another (e.g. fans of different sports teams). At the same time, people appear to display a higher degree of counter-empathy towards members of a competitive group. There's a lot here that can be useful to us, but one of the really fascinating parts is that this intergroup bias even seems to appear in groups that have only just formed. Participants in some of these experiments had just met and had no history with one another whatsoever. The only condition for bias to appear seems to be that the groups find themselves in a competitive situation.
Through carefully developing intergroup conflict we can make the player more likely to care about gameplay scenarios than they were before. By applying it we could strengthen the feelings of empathy we are aiming for, as they are what will have players coming back for more. Likewise, we get to use counter-empathy to our advantage, too. There are many examples of games that have introduced foils for the player, rivals to fight against, or ‘nemeses' if you will. Using mechanics of intergroup conflict between equal teams with recurring characters should be even more effective at creating relatedness. Intergroup bias provides a powerful framing-device to create more emotionally impactful experiences in the short term and loyalty in the long term. Conflict between our friends and rival NPCs should build sustained engagement through a push and pull of emotions.
This should all sound very good. But as with any good-sounding claim in any context ever, there remains one crucial question: Does it actually work? Short answer: Yes! Long answer: Let's talk a bit about my data...
To see if this would be applicable to game design, we recreated a previous psychological study from 2014 in a gaming context. We made participants aware of the video game setting (participants in the 2014 study believed they were competing with real people), both through text as well as cartoony 2D profile pictures for the NPCs. With our research prototype complete, we paid 116 people to play it via Amazon's online tool Mechanical Turk. All participants played through the research prototype and had their empathetic reactions recorded via a questionnaire.
We used 2D representations of our NPCs to make their fictionality clear to the player
Between gameplay sessions we asked participants how they felt about certain scenarios that happened to the NPCs. These were either positive (e.g. "Monica ate a really good sandwich.") or negative (e.g. "Andrew stepped in dog poo.") and happened to members of both teams in equal numbers. Participants indicated their feelings of empathy on sliders, which asked how good and how bad they felt about the NPCs' experiences. We predicted that participants would react to the experiences of fictional characters similar to what was observed in regard to real people in the previous study.
We used 2D representations of our NPCs to make their fictionality clear to the player
As we expected, our participants felt more empathy towards their own team than they did towards opponent-NPCs, for whom they in turn had more counter-empathy. Not only do these results mirror observations from previous research that bias will appear within novel groups, they also show that this is the case in a fictional and game-like setting (i.e. with NPCs, not real people). This shows that intergroup conflict could at least be a useful tool to enhance emotional reactions towards both teams, as it clearly appears in regard to NPCs in a game. What also should be pointed out is that we were able to observe an empathy bias around text-based, very low impact scenarios. It is conceivable that player reactions to actually seeing a friend or foe being shot in the face with a shotgun should be more pronounced.â€‹
Differences in participant ratings about their empathetic reactions (different scales used for better readability)
But wait, there is more. We even found a few factors that seemed to modulate the strength of intergroup bias in players. The first one is, that the level of bias seemed to get stronger as the experiment went on. The differences between the reported reactions grew larger between the first and second round of questions. This makes sense as emotions should get stronger over time, while relationships begin to develop.
Yet, we also saw that player performance (i.e. their win-rate) seemed to affect their feelings. Players had the possibility to win up to two rounds of the research game, before reporting their reactions for a second time. All empathy ratings seemed to decline the better players did (most at 0 wins, the least at 2 wins).
Counter-empathetic reactions did not follow that trend. Instead, they seemed to peak when the gameplay experience was the most contested at 1 win and 1 loss. In both cases, the differences in empathy between the player- and opponent-group remained constant.
Differences in participant ratings about their empathetic reactions by number of wins (different scales used for better readability)
Our experiment clearly confirmed our expectations that intergroup bias will appear in a gaming context, as it would in the real world. It seems likely that enhanced empathy through a targeted use of bias could lead to stronger relationships with NPCs and as a result, improve player-enjoyment and -retention. Just introducing intergroup conflict could lead to a higher degree of empathy for player NPCs per default. Likewise, having recurring NPCs compete in groups of equal standing would add meaningful rivalries into the mix. Finally, we not only know that it works, but also how we could modulate the effects. Having control over data about factors like the time of exposure and player performance in our game should make the efficacy of bias as a design tool easier to test for.â€‹
If you are interested in exploring how intergroup empathy bias could be useful for your game, here is a short check-list of things to keep in mind:
Jonas Pastoors is a game designer currently looking for new opportunities in NPC-design. His most recent projects include Iron Harvest and Maze Slaughter.
To get in touch, visit:
Follow them on Twitter:
I would like to thank Thomas Buijtenweg and Mata Haggis-Burridge for their valuable input and review of this article.