Selective listening happens before listening begins. You reach for some voices and reflexively discount others — based on accent, identity, vocabulary, gender, role, age, or who said it first. From the inside, it doesn't feel like discounting. It feels like accurate filtering. But the same words from a different speaker would have landed. This is the listening expression of Force 04 · Bias, and it shapes every conversation before a single sentence is exchanged.
The core distinction
The other three impostors are about how you listen. Selective listening is about who you listen to. They are different problems. You can be excellent at the long pause and the mirror-back and still apply both unevenly — generous attention for the people whose voices you've already coded as credible, and a faster filter for everyone else. The faster filter is the impostor.
The reason this is so consequential is that listening is the first peacemaking act. If listening is unequal at the level of who gets received, every other practice in this course will be unequal too. The dignity check (Force 04) and selective listening are the same problem expressed in two domains: who gets to count as fully human in your mind, and whose words get to count as worth hearing.
The research
Accent reduces perceived credibility — independent of content.
A widely replicated 2010 study by Lev-Ari and Keysar found that the same factual statement was rated as less truthful when delivered by a non-native speaker than by a native speaker — and the effect was stronger when the accent was heavier. The participants were not aware of doing this. They reported listening fairly to both. The accent did the work in the background. Subsequent research has documented similar effects for regional accents within a language, for vocal fry, and for vocabulary registers associated with class or education level.
Source
Lev-Ari & Keysar (2010), "Why don't we believe non-native speakers? The influence of accent on credibility".
Who gets interrupted is patterned — and the listener is usually the one doing it.
A long line of conversation-analysis research, beginning with West and Zimmerman's 1975 work and replicated many times since, documents that interruptions are not evenly distributed. Across studies, women are interrupted more than men, lower-status speakers are interrupted more than higher-status ones, and members of marginalized groups are interrupted more than members of dominant groups. Interruption is selective listening in its loudest form — the listener has decided this speaker's sentence is not worth completing.
Source
West & Zimmerman (1975), foundational; see also Hancock & Rubin (2015) and ongoing replication.
Selective exposure produces selective hearing — even before bias-driven evaluation kicks in.
Decades of selective-exposure research show that people spend more time with information from sources that align with their existing views, and less time with information from sources they perceive as misaligned. Stroud's 2008 work and many follow-on studies document the effect across media, conversation partners, and political contexts. The downstream consequence: by the time a speaker from an "outgroup source" is in front of you, your brain has already had less practice receiving them.
Source
The same words are evaluated differently depending on who says them.
A series of studies on "source effects" in persuasion research demonstrates that identical content is rated as more or less persuasive, more or less reasonable, more or less worth engaging with, depending on the identity attributed to the speaker. This is not always negative bias — sometimes the speaker is over-credited because of prestige. Either way, the content has not actually been heard. What has been heard is the source.
Source
Hovland & Weiss (1951), foundational; broad subsequent literature on source credibility and persuasion.
Contact and individuation reduce the effect.
The same research base that supports the dignity check supports the counter-practice here. Pettigrew and Tropp's 2006 meta-analysis (515 studies, 250,000+ participants) shows that contact under good conditions reduces bias in 94% of studies. Individuation research — actively perceiving members of an outgroup as distinct individuals rather than category representatives — measurably reduces the selective-listening effect. Listening practiced across difference, over time, with curiosity, is the work that loosens the filter.
Source
Pettigrew & Tropp (2006), "A meta-analytic test of intergroup contact theory".
Catching yourself doing it
The practical correction is to deliberately listen harder to the voices your brain has already coded as less credible — not as politeness, but as repair of the filter.
The counter-practice
Mirror-back is the answer to selective listening. It is the practice that requires you to have actually received the speaker's words — because you have to say them back. A selective listener cannot mirror back well. The discipline forces the filter to slow down.
Counter-practice · Week 03 · Practice 02
Before responding to what someone said, say back what you heard them say. Not paraphrasing to sound smart — actually showing them you got it. The practice works equally on every speaker — which is exactly what selective listening doesn't.
What it usually protects
Selective listening is often defended as efficiency: "I can't listen to everyone the same way. I have to prioritize." That is partially true. Time is finite. Some sources are genuinely more reliable than others. The problem is that the priority-setting almost always tracks lines of power, identity, and familiarity, not lines of actual credibility — and the listener doesn't notice.
The corrective is not to listen to everyone with equal weight regardless of evidence. It is to notice your filter — and to occasionally slow it down, especially when the conversation matters. The people most likely to teach you something new are usually not the people your filter reaches for. They are the people your filter quietly demotes.
Why this impostor matters
The impostor, in one sentence
This is not a moral failing. It is what brains do — a downstream consequence of bias (Force 04 from Week 02), running in real time on every voice in the room. The selective filter is automatic, fast, and silent. From the inside it feels like you are listening fairly. From the data, you are not.
The work is not to become an undifferentiated listener — that would not even be coherent. The work is to notice whose voice your attention runs from, and to occasionally do the harder thing of staying. The dignity check (Force 04) and selective listening are the same problem in two domains: who counts as fully human in your mind, and whose words count as worth fully hearing. Both questions get answered, one conversation at a time.