The culture groups in CK2 are (with one or two exceptions^*) based on linguistic affinity, e.g. there's a "North Germanic" group for cultures speaking North Germanic languages, a "Latin" group for cultures speaking Romance languages, and so on. That's a sensible choice: there's really no objective way of grouping together "cultures" (if indeed there is such a thing) otherwise.
Sogdian was indeed an Iranian language, a close relative of Persian spoken in and around Samarkand, and so one of the lingua francas of the medieval Silk Road.
The Tocharians' placement makes less sense as Tocharian wasn't an Iranian language. It's an odd language which belongs to its own family within the Indo-European languages. It was probably the last remnant of a very ancient migration of Indo-European speakers into Central Asia, which was later eclipsed by the spread of Indo-Iranian and Turkic languages. Presumably however putting this quirk of historical linguistics into the game would've had a disproportionate impact, so Paradox just lumped them in with the dominant Indo-European culture group in Central Asia (Iranian).
^* Okay, I can't resist pointing these out:
"Iberian" mixes Romance (Castillian, Catalan & Portuguese), Germanic (Visigothic) and non-Indo-European (Basque) speakers.
"Byzantine" encompasses four essentially unrelated languages, one of which isn't even Indo-European (Georgian).
Hungarian, despite being a Finno-Ugric language, is in its own group, "Magyar" – which is simply the name of Hungarian in Hungarian.
"Altaic" mixes Turkic and Mongolic languages. These notoriously do not mix. Pedantism aside, I think this was a poor choice by Paradox. It lumps all the (non-Iranian) steppe cultures together and means the Turkic cultures have warm feelings for the Mongols when they show up, which they really shouldn't. The Mongols were invaders from the eastern steppe whose impact on Turkic Central Asia was no less catastrophic than it was on Europe, Persia or China. It also introduces the inconsistency of the Turkic cultures liking their long-term neighbours with Iranian culture less than their Mongol conquerors (and the Mongols liking their Turkic-speaking steppe subjects more than their Iranian-speaking steppe subjects).
Vlachs (AKA Romanians and pals) speak a Romance language, but CK2 considers them South Slavic.
"East African" is an understandable catch-all. Somali is a Cushitic (Afro-Asiatic) language. Nubian is (probably) Nilo-Saharan. Nearly a hundred different languages are spoken in Ethiopia, belonging to various branches of the Afro-Asiatic family with some Nilo-Saharan thrown into the mix too.
And I'm sure I missed some.
Just to clarify on OP's question, Tocharians and Sogdians are their own "cultures" on the map, but are considered part of the Iranian Culture Group. This group also contains Persians, Kurds, Afghans and Balochs. Culture groups in game contain cultures that are considered closer to others within their group than to cultures outside it.
Sogdian is definitely classified under the category of Iranian. It survives today in the form of Yaghnnobi in modern Tajikistan. Kurdish, Pashto, Balochi, and Persian are also Iranian languages.
Tocharian, however, is not an Iranian language. Like the Iranian languages (as well the Romance and Germanic languages) it is part of the Indo-European language family. However, it is a unique and extinct branch that was historically spoken in the Tarim Basin in what is now Xinjiang, China. It was replaced by Uyghur, a Turkic language, during the 9th century CE. Sogdian and other Indo-European languages that were historically spoken in Central Asia were also, with exceptions, replaced by Turkic languages.